Decals

Use the X and Y Repeat fields in the Decal Image Shortcut Properties It's not exactly automatic, as it will only tile across the area of your decal, so the decal has to be placed properly for starters. You'll also have to play with the number of repeats to get the proper scale for your tile.

I had been using the cell turbulence thing for irregular color/spec maps and it works really nice. I used plated with 2 or 3 octaves. The cellturb combiner is the best.

Any types of wet looks are accomplished with effective use of specularity controls and mapping. For example, if the specular color were left at the default white setting, the model Alan made would probably still look wet, but it might not look like wet alien skin. By setting a specular color that is similar to the main texture color, it has a totally different effect.

The bump map also helps to break up the specular area, which enhances the wet/slimy effect. Notice that the same specular color is present in the back of the head, but because the surface doesn't have any bumpiness, it doesn't quite have that slimy appearance that the head does.

There's also the option of using a specularity map to vary the specular highlights across the model's surface.

Justin Barrett

Yes it can! On of the very powerful things about A:M texturing is that we are not limited to just the major orthographic views to apply decals. (And there are suposed "Big Boy" 3D apps that restrict you in just that way).

You cannot rotate the image itself when applying it, but you can position the view of your model in any position from any view from almost any window for each decal. It is very flexible. I usually apply decals in an action window, but a Pose or modeling window works fine too.

The decal containers are fantastic as well. (I only wish that each stamp had position properties panel showing the numbers like the Decal properties panel itself).

And another cool thing is A:M's Invarient procedural textures. Once you apply them to your model they stick to the surface of the model and deform with it, making them wonderful for character animation.

NO other 3D app has this feature. :) With other apps, when you deform the model the surface moves through the textures local or global 3D space. Some other apps can "Bake" the procedural textures to the surface of the model and make them into map(s) with that texture, but then you loose the infinite resolution of an algorithmic procedural that never pixelates when you get to close.

I've also had good results with 400 oversample in rendering animations with high frequency. It killed almost all the flicker/"sizzle- pop". There is a signifigant improvement over 200% oversample, or no oversample at all.

Glen Crowell

I discovered something very useful that others might like to know about. This is more of a tip than anything. Most of the time I can't find any textures that match the subject I am working with, and I don't want to have to take an hour to create one. So what you can do is use an image that has the same sort of surface, apply it as an image, and then stick it at 50% under the properties, then select the same thing again and give it a colour. The colour and texture then can mix to create an okay looking object.

As an example, I had a pear that I wanted to texture and had the perfect pear texture for it, but I couldn't seem to get it to match up and look nice as a decal, so I just put it on as an image. Not all the patches were the same size so you could see a bit of squashing and stretching, and it was obvious that it was a repeating texture. So I went under the image options and set it 50. Then I made the pear green under the colour settings. That was just enough so that the artifacts in the texture disappeared but the pear still had the proper look on the surface. I've done this with a bunch of other stuff and it seems to work quite satisfactory.

This may be common knowledge but I just thought it might be able to help someone out, who like me doesn't have a great image editor, and doesn't have the money to buy a good texture CD.

Matt Jones

The best solution I could find is to render with oversampling at 400%. This takes care of even single pixel contrast in bitmap textures. But render times will suffer ...... :-(

Sotiris Gougousis

If you have a big texture map, try blurring the textures a bit. Render final with filter turn on. TV and Video have interlace thing so if you're outputing it this medium, never use single pixel lines ar dots. Try not to have dense lines to avoid moire pattern effect.

-sine

One thing to check is the value of your bump maps. If the value is to high it can cause the shimmering.

If you back off the value and the shimmer goes away, you will probably need to build the detail into the model.

mike hovland

One thing to remember is to make sure there are no patches lying on top of each other once you have flattened it. In other words, all patches should be visible to you. You can turn on shaded mode if it's not clear to see if some might be hidden. I believe that if part of a patch is hidden it won't take a decal.

Good point. When flattening in a pose, I find that the easiest way to locate these patches is to scrub the pose slider back and forth to see how A:M flattened the model. If CPs are overlapping, you can scrub the pose slider to about 90 to select one of them and move the slider back to 100, with the CP still selected, and move it where you want it.

Yes it is. Apply both decals as color in the same decal container. Make a new action, drag the shortcut of each decal image onto the action PWS, for each decal shortcut (now in the action) check the percentage box to gain access to the channel. Adjust percentage to your liking.

Now that I think of it, you can do the same thing with two different decal containers.

Jeff C.

Here's a cool tip for displacement maps...

Add your displacemnt map as a bump map on top of the original displacement map. Other programs (such as Maya) REQUIRE a bump map to be associated with any displacement map in order to determine how the displaced surface's normals should be oriented at render time. I tried this with a low density patch object in Hash with great results.

A highly detailed bump/displacement map will give you general surface displacement in the large areas of dark and light on the map (Imagine if you squint your eyes to blur out your map, those dark and light areas will be bulged out from the surface.), the smaller details in the map will give the illusion of displacement when used as your bump map. I created the muscles in a low-patch count character's chest using this technique and all I had to model was less than a half dozen patches to map the displacemnt/bump map onto.

-Wesley Grandmont III

Yup, see page 332 of the newest manual... Manual says drag the decal shortcut from the images folder to the "choregraphy action" under the model in the cho section. What I found was you need to drag the shortcut of the applied decal from the model's decal, image folder, then it works after you put a check mark next to "Frame" in the newly created Decal-shortcut to (filename) prop panel. That creates a new frame channel that allows you to type in the value you want and when you want it.

As a side note, no change was noticed with progressive render, but in final render the decals updated and displayed correctly. I have an intel box. The channels viewer for the frames also displayed goofy data -- what was listed as one (1) on the viewer gave a value of 30 on the data point, which was confusing for a bit.

Any hoo, it works for me by directly typing in the value I want and rendered in final mode.

Hope this helps make it work for you too,

Jeff Cantin

I'd use layers of projection maps. One for color, one diffuse, one bump. (make a seamless pattern and repeat as necessary). Any graffiti on the wall can be added as overlay decal or projected color maps. (image with alpha channel)

-sine

Based on my experience, building the wall from scratch calls for a lot of tedious spline wrangling and the number of cp's add up quickly. For example, using my modelling technique (all corners bevelled) a 4 block x 36 block wall generated a 550kb .mdl file. But it can pay dividends. Once I had my patch wall, I was able to use it to generate colour and bump decals that I was quite satisfied with.

Basically you make the blocks faces one group and the mortar another. Then apply appropriate procedural materials that are sufficiently random that they can be tiled. Do a screen render and capture that to your paint program. This is your colour decal.

Then make a copy of the wall and apply a gradient material that transitions from black to noisy white as you go from the base of the mortar to the tops of the block faces. Do another screen render and capture. This is your bump decal.

-Rodger Reynolds

Again, this is one that I'm sure most of you have found already, but I found it so painfully, evilly useful, I have to share it.

When you're ready to decal, screen grab your model in wireframe. Import it into photoshop, crop it, then before you do anything, save it back out as a tga.

Flip back to AM, create a decal, and import the image you just saved. Because you're placing a direct copy of the wireframe over the physical wireframe, you should be able to line up the image much faster than lining up a vague colour map or bump map.

Once you've done that, either turn off, or delete the image from the decal entirely. The position information for the decal will remain. Back in photoshop, create your colour/bump/spec/diffuse maps in seperate layers over the original wireframe. Then for each layer, make sure it's the only one visible, and choose 'save a copy' from the file menu, and save as a tga. This way, you don't have to save a backup, and then flatten for each layer to get them out of photoshop and into AM.

Once you have all your tga's ready, right click on your decal, and choose 'new image'. Find the relevant image, change it's type to match, and hey presto, it's instantly applied and positioned. You don't have to copy top-down-left-right values for every seperate decal, and you keep your maps nicely organised in the PWS.

Call me mr. big gibbering idiot, but I never fully realised how useful it could be to have multiple images within one decal. Great timesaving stuff.

-Matt Estela

Copy your image into alpha channel (#4), then fill in your image with all black. Save this as a 32 bit targa and apply in A:M.

The alpha channel will only show your black letters on your model without a black halo ring or aliased pixels from just using the transparent color option in the image properties page.

Glen Crowell

Generally I apply the texture as one whole map. i flatten the character's head in action and capture the mesh with paintshop pro. i use this wireframe mesh as a guide to create the color map. i then turned the color map into a grayscale image and darkened the creases and lightened the areas that would be high. this grayscale image is then applied as a bump map. i then took the same garyscale image and removed all areas that wouldn't reflect light and used this image as a reflection map for subtle shiny parts. this same image was used as a deffuse map as well but set to around 40%.

when creating the color map i only work on one side of the face and apply it to the character and test render. i return to photoshop make some adjustments and save the map and test render again (in a temp chorography with lighting) i repeat this process until the texture looks good. when i think its ok i mirror the other half and add different deatils to different areas but this of course is based on how much time i want to put into a character which in this case i was just fooling around.

frank

The point I wanted to make (other that re-introducing mesa to yousa (arrggghhh!!!!! Jar Jar has infected my keyboard!)) is bump maps are best handled as multiple maps for various levels, or magnitudes of surface detail.

I typically have a the following maps:

1. A color map (what some apps. call a diffuse map, as it affects the diffuse, or bounced color, whereas AM considers diffuse map to be more of a luminance map)
2. several bump maps. In Darth's example I've got:
   1. a skin texture map - this covers the basic crackly look that skin has
   2. a wrinkle map - this covers the deeper, more detailed wrinkles in places like joints, foreheads, lips, eyes, etc.
   3. a lump map, this covers things like fat and muscle detail below the skin
3. I rarely use what AM calls diffuse mapping, as I'd rather adjust coloration variation in the color map itself. For what it's worth.

Here's a bit on how I do bump maps. I use Photoshop, so that's how the terminology relates. All bump maps are drawn/created with a background of RGB:128,128,128. This way you can have bumps as well as depressions in the same map. White=bump, black=crease/wrinkle.

skin map is just a heavy noise layer, blurred, then apply the crystallize filter to get the "cellular" look, then a find edges, or glowing edges, to make the little creases between the cellular areas. Play with blurring and fading/transparency settings. Do as much as you can on individual layers so you have more control over the final composition of the bump map.

Wrinkle map is just a series of layers with positive and negative bumpage (black and white lines/areas)

Wrinkles are the key to realistic textures as far as I'm concerned. The best way to make wrinkles look real that I've found is to have a tablet, choose a smallish brush, have pressure adjust size, opacity, and maybe even color. Draw some basic wrinkles, then do a blur filter. Draw some more wrinkles over those wrinkles, and then blur it again. Repeat this process a bunch of times.

What your doing when you draw/blur over and over is adding several "levels" of detail. One problem I see with a lot of bump maps used as wrinkle maps is that they are very one dimensional. What I mean is that there are areas of wrinkles, and there are areas of no wrinkles. On a real body, there is almost no area that has no wrinkles, and not all wrinkles are the same "magnitude". By doing the draw/blur thing over and over, you can get some really good looking wrinkles/textures on the skin that vary in absolute as well as relative magnitude.

The lump map, is essentially the same technique as the wrinkle maps, but I use the airbrush tool to make larger areas for the bumps/lumps.

*** A final technique to get good bump maps is to apply the despeckle filter a few times to "even" out the bump map gradients, so you don't get rough looking bumps. Use the histogram in Photoshop to see if you have a nice even distribution of grays. If the histogram is rough or bumpy, your bump maps will be also.

Now you have three layers to use as decals, and therefore you can choose the amount for each, For example I think my wrinkle layer is an amount of 1500, skin layer of around 800, and the lump layer of around 500.

You can use more extreme gray ranges in your bump map rather than jacking up the bump map setting to level like 1000+ in AM, but I think that smoother bump-maps, with higher settings in AM make nicer bumps.

So, thus ends my blather, and I go back to lurking...

Todd Shafer

Here is a method for getting rid of moire patterns that I used in my print graphic days.

Scan in at 300 ppi (any larger and the following steps don't usually work to well as the individual, printed, dots encompass more than a few pixels each). Preform the following filters in this order to the raw 300ppi scanned image:

• Despeckle Twice
• Gaussian blur @ .7 pixel radius
• Median 1 pixel radius (sometimes 2 pixel radius)

(and after a little levels adjustment and whatnot, I usually do an unsharp mask at like 150% 1.8 pixels, and 4 threshold levels.)

Scanning at a higher rez than what you ultimately need is good because each time you do something to your photo, it reduces the information left. But higher that 300ppi on anything other than a real photo doesn't really do it for me. Here's why:

What I like to do is scan in at at 300 ppi as it roughly encompasses most linescreens (such as 133 to like 175 linescreen), which is what most stuff is printed at. depending on if you use the 2Xlpi or 1.5XLpi method for achieving like a 150lpi photo. i.e. 150lpi X 1.5 = 225ppi or 150lpi X 2 = 300ppi. more is better, but not at values over 300ppi.

If I were to scan a printed piece at 600ppi, then the procedures for reducing noise in the scan become much less effective. The "noise" of the moire becomes less of noise than a chunk of pixels per dot at higher resolutions. That's pretty much why I like to scan printed material in always at %100 scaling and 300ppi regardless. If the printed piece doesn't work with my method, I find another printed piece that will.

so unless something was printed at 300lpi, (which would make the rosettes much smaller and more 2X pixel size of a 600ppi scanned printed image) I would not go above 300ppi

-Armando Afre

Create a new action with your model. Drag the Shortcut to your image sequence (from the Decals folder) into the action. Look at it's Properties. There will be a "Frame" channel available. Check it to create the channel.

Advance to the frame after the last frame of your animated decal, and look at the propertis for the Frame channel you just created. Set the "Value" to 00:00:01 and then check "Repeat."

That otter do it. It's important that you set the value *before* checking Repeat, cause otherwise you get a "division by zero" error (at least I do).

David Fontes

To animate a decal percent, make a new action and drag the shortcut of the image used for the decal into the action.

Jeff C.

Here is a technique for decalling that simply -everybody- ought to be using nowadays.

To decal, you'll want to create a pose that stretches and flattens out the area you're going to decal. Then hide everything but those points, zoom in, and take a screenshot.

My -first- action (after saving the screenshot from Photoshop) is to import that screenshot into AM, make a decal with it, line it up with the on-screen view perfectly, and stamp it.

Then you can remove the one screen-shot image from the images container, and replace it with the decals you create over the screenshot. But the Stamp will still be exactly the same mapping, so at least you will know that your decal pixels are going on the model in exactly the same way you are drawing them in your paint programs.

-- Tony

If you've got a texture map that's too regularly spaced (like scales straight out of DarkTree), try the following:

• Create an 8x8 patch grid
• Decal the texture directly onto the grid.
• Move around the internal control points to change the proportioning, or just to break up the regularity of the decal.
• Render and use the rendered (i.e. post deformation) output as your new decal.

You can use this method to process a whole batch of textures through the identical transforms: just swap images into and out of the Decal's "Image" folder, and render out each item after its deformation. Fun stuff, eh?

-- Tony

In the past few weeks there have been a few postings about some patches not taking a decal, I did some experimenting on this to see if I could figure out what would cause this. Here is what I learned.

Although I like to keep track of my normals, they didn't seem to make a difference when decaling. I reversed some as a test, the decal stuck like it should. I tested color, bump, cookie cut, etc... the normals direction did not affect the decal nor the bump direction. Special case is "Weathering" which *is* normals aware.

5 point patches and hooks both accepted the decal just fine.

Flatten... Here's where I found I could make portions of decals *not* stick. Overlapping splines was the main cause. To understand what I mean, take a sphere and hide the back half, then in a pose window, flatten by hand and take one longitudinal spline and pull it over the one next to it, apply your decal and render in the model window to see the effect. Some patches will not have the decal applied.

Think nostrils on a face, where the nostrils attach to the face could be behind the rest of the nose splinage... If you just applied a decal from the front view that "behind" part wouldn't get decaled. When you flatten and don't tweak that area, that patch will not get the decal data either.

I hope this data is useful, if it happens to solve your problems, would you drop me a note?

Thanks in advance,

Jeff Cantin

I've noticed that sometimes the gamma setting will change itself. If you select the decal and check the properties palette, there's a gamma adjustment there. My guess is that's your problem.

John Koester

It doesnt exactly change itself. I can't remember where but you can set the gamma of a tga in photoshop. If photoshop sets it to anything other that 1 it wont show in photoshop but it will show in AM.

eric

If you go to the Property Panel while you have the Rotoscope selected, it has a dropdown list that defines which window the rotoscope should appear in. Set it to Right or Left, and the item should appear in the corresponding side view.

--Tony

Decaling with alpha channels:

Alpha channels are the way to go when applying color map decals. You can layer as many maps on top of an object as you like, which is very nice for builing up subtle surface detail without enormous files sizes on the image map.

If there is an alpha channel available the background color option is greyed out completely. This means it is using the alpha channel for the transparency of the map. Alphas can be painted by hand or created using selection tools in Photoshop or Painter. Make sure the gamma is what you want (usually 1.0).

If you have an alpha channel on a color image, the decal will show through to the underlying color (rather than making the object transparent). When using color maps, with or without alpha, the last color map you apply renders on top. You can make a grayscale texture as a transparency map to make an object transparent. By aligning the two (you may have to do some manual typing), you can create a texture object that fades to transparent wherever you like.

First thing to do is give it a try. Yes, Hash allows alpha channel, yes it can be a single image, yes you can vary the amount of color via the prop panel percentage, yes you can do as you proposed... yes, yes, and yes... :-)

Jeff C.

Yep there is a way to flip your decals. Either in X or Y. you just swich max min values in the decal properties. And the decal is flipped. you could also do it with your mouse but that wouldn´t be as exact.

Sincerelly Peter Tomsson

No hash can only have rotoscopes in the cardinal viewpoints (right left front.. etc.(except of course in the choreography window which also uses a camera rotoscope.) as a work around I would suggest applying the image you want to have rotoscopes at 45 degrees is to put the rotoscope onto a plane flat patch and place this patch at 45 degrees behind your model. assign a bone to the rotoscope patch and set ot to shaded mode. make sure you have the show decals in the rendering preferences panel turned on. Then set your birds eye view to the appropriate angle

Note this is not a perfect solution and you may have to switch the draw state of the bone back and forth from shaded to wireframe to keep things working properly. but if you really want a 45 degree rotoscope without rotating the whole model in the front view then this the best way I can think to accomplish the task.

-David Rogers

All you need to do is add it to the images folder in the decal you want to add it to.

Brian "balistic" Prince

I've had problem like this where a couple patches wouldnt show my decal. what i did was i deleted the .ava file of the same name (it will be in the same folder) and then it rendered fine. im not exactly sure but i think the .ava file stores all the info about textures and models so when you do quickshade it doesnt take as long to load the texture maps.

Eric Litton

No! I wouldn't do that. Take a head for example. You have a face that going to get lots of maps, two sides of the head that will include the ears, the jaw and the sides of the head, and you also have the back of the head. Each of these will be mapped.

What I do is hide everything except what I'm decaling. For the face, I hide everything but the face. Then create a new pose (which you will never use again -- it's just temporary). Read the manual for how to setup the rotational axis before flattening, and then Flatten.

Once flattened, I do a quickrender, take a screencapture, paint the map, apply, and boom! Done.

Close, but I don't think that's exactly it. Flattening works like a cylinder. The axis of the cylinder is the Z axis. Picture a cylinder running along the Z axis. When the Flatten occurs, the cylinder is unraveled. I think the point where it unravels is at the top of the cylinder, up in the positive Y direction.

Now, I'd say its pretty common to build characters facing forward (i.e. facing in the Z direction [are you with me?]). The imaginary Flatten cylinder is pointing straight out of their face. We need to rotate the cylinder so it runs parallel with the Y axis.

This is done by selecting the points you want to flatten (you should be in the Pose now and have only the points you want to work with showing) [select points in Pose using Muscle] -- and then hitting R as though you were going to rotate them.

Instead of rotating the points, we're going to rotate the control handle at the very center of the selection. There are three handles. One of them points straight up (Y axis). Rotate this back and watch the blue Z axis handle rotate up into the Y axis position. Do not deselect.

This basically rotates the Flatten cylinder from the Z axis to the Y axis.

While all the points are still selected and the Z axis is pointed straight up, right-click and select Flatten.

If you add a matched set of maps, color, bump, and transparancy for example, and then need to tweak the transparancy map, here is a tip that worked for me. V6.1

Have your transparancy map in your image editor, switch back to AM98 and from the PWS, right click on the decal container (red star icon) and change the "decal to display" to the trans map, then right click again on the container and choose "position." You should see the map and your mesh underneath it. Do a screen grab (on a PC, whack the "Print Scrn SysRq" key). Open a new image window in the image editor and paste in the screen grab, you can crop it to just display the area of your decal.

Using the screen grab as a refference, adjust your transparancy map & save it, go back to AM98 and do a quick render to force AM98 to update the map to the current file changes and to see your results. Repeat as often as needed or adjust your map from your own memory cells...

This should work for tweaking any set of maps.

Jeff C.

You have to load all your decals into the project, so that they're accesible from the list, then once you've started a new container (decal item under the model) and applied the first map expand it down to where the name of the decal is listed, go up one level and right click and it will allow you to add more decals with the same stamp bounds to the container. If it looks like it isn't displaying all the decals in the container properly (i.e color with no bump or grey decal when there's also a color map in the container) save the project, close the program and relaunch and everything should come back in properly.

Matt Andersen

Yeah, a lot of vid cards don't display positioning transparency properly. Save your decals with an alpha channel and turn off the diplay transparent box in the decal's general tab.

Matt Andersen

Well, I happened upon a nice new trick for increasing render speed in specific situations, so I figured I'd share.

Basic principle: Do -not- invoke the raytracer if you don't need to.

Specific instance: I have a Hair map which covers a large area of splines. Around the edges, it has stray hairs and somewhat ragged edges, and like that. So I used a Cookie-Cut map with a transparency channel, and everything was working great. Except, of course, that it was rendering in 1:45/frame for an 80x60 closeup. Not so happy, but understandable given the detail of the model.

However, I had a wierd thought while out running (that happens). Most of the mesh of that hair does not have any transparency effects. Only the edges actually need to be cookiecutters. So, when I got home, I created two decals: one a cookie cutter and one a plain color. Put them in the same location. I also created two new Groups. One grouped only the edge patches that needed transparency. The other grouped the rest of the patches of the hair. Note that both include the "border splines" which seperate those patches.

Then I mapped the cookie-cutter only to the edges, and the color only to the interior patches. I rendered, and everything looked exactly the way it had previously, but the render time dropped from 1:45 to 0:25!!! I was, to say the least, pleased :)

This was, of course, an extreme example. I was going closeup on the face, which put a lot of the hair in frame, and the decal map is disgustingly large, and there are shadows and multiple light-sources, and so on and so forth. Still, that's a pretty buffed out speed increase from such a little modification.

Well, just thought I'd share. I hope this helps out somebody!

-- Tony

I use both Photoshop and Painter, but I prefer Photoshop....

I don't know if Painter does the same thing, but once a color map is made in Photoshop and saved, selecting Image, Mode, Grayscale will convert it to grayscale, then select Image, Adjust, Invert will usually create a perfect bump/specularity/transparency map. This is really quick and easy, and then each of the images can be added to a decal container and stamped on in one shot.

Jeff, Hash, Inc.

AM doesn't need it any longer. A couple of betas ago they removed it. It should automatically load the updated texture if a newer one is available.

I happened upon a sort of nifty texturing method recently. This may be old hat to some people, but if you haven't seen it, then it's new to you!

What I wanted to do was to rough in the colors on a face very quickly, prior to doing any detail work. It also turns out that this method helped me to play around with the balance of colors (for example, on my first attempt I made the lips much too red) without the effort of major reworking.

As per normal practice, I made a group with only the patches I wanted to texture (MAP Face), and a pose in which I manually flattened and spaced those patches (also MAP Face). I hid all the other patches, made the pose a big-window view, took a screenshot, and imported it into Photoshop.

In Photoshop, I looked through the various R, G and B channels, to find the one where the splines had the most contrast from their background. I copied this into a new channel, and used Levels to make the contrast total (i.e. the background is dead black and the splines and CPs are bright white). I ended up with a channel which had very sharply aliased edges between the CPs and the surfaces of the patches. Save this as "Face Paint by Numbers". Do not resize or resample it, or you will lose the sharpness of the edges: the non-aliased edges are essential to making the technique work.

Now I saved it again, as "Face Color". I went into the alpha channel and painted it all white. Then painted black around the edges of the selection, and blurred it. This alpha channel will give you a gentle fade into whatever your model's basic skin-tone is. Except, of course, that the RGB channel is still showing a screenshot. We'll be fixing that shortly.

Go back to the Paint-by-numbers map. Go to channel 4. Use the magic wand, and click inside a patch's black area. This will highlight that patch. Use 'Select->Similar' to select all black areas. Click on the RGB channel, and fill those areas with your model's basic skin tone. Select All in the RGB channel, copy and paste into the RGB channel of the Face Color.

Save the Face Color and go into AM98, and apply it to the face spline. Bring up a nice choreography view of your character, and render the face. You will see normal skin tone, interrupted by a projection of your splines onto the surface. This is your guide.

Now comes that painting by numbers. Go back into the Face PBN image. To select a patch, just go to channel 4 and use the Magic wand. Use 'Shift-click' with the magic wand to select multiple patches. Then, to paint those patches uniformly, switch to the RGB channel and just use Fill, with whatever color you feel like.

Any time you want to see the effects of your painting on your model, just copy the RGB channel from the paint-by-numbers into the 'Face Color' image, and save it. When you render in AM98, the program will notice that the image has changed, and reload it.

If you decide that a color you used was just too garish, all you need to do is reselect the patches and fill with some other color. You can get a fairly tight cycle of modifying your image and examining the results on the model.

You can also use the same technique to create any other types of maps: specularity, bump, diffuse, whatever. There are a few tricks dealing with converting plain color into a degree of bumpiness (use the color-intensity channel as a selection, then apply a gaussian noise filter to a new layer using that selection), but they're fairly straightforward.

Then, once you've got all of your maps working in the rough, you can just take them into Painter (or whatever you use for detailing) as templates, and fill in the sharp edges, and whatever other detailing you want to do.

Again, if this is obvious to you then I apologize for wasting your time. If it's new and exciting, then I'm glad! Hope everyone has fun!

-- Tony

I usually just set a threshhold value so the splines appear black, everything white. I select the black splines only and copy it to a new layer in a different Photoshop file. I then leave Spline Layer like 30% visible, but untouchable, and begin painting and such in the Background Layer.

This is all done in photoshop. If I need to see the splines more or less, I just adjust the opacity percentage of the spline layer. Because I've only selected the black splines to the new layer, all the "white" areas of the spline layer are completely transparent regardless of the spline opacity.

-Armando Afre

You probably have an alpha channel in your picture. Hash will not let you change the transparent color when an alpha channel exists. Either delete the channel in your paint program or save it as a pcx.

Chris

There are a few techniques that you could use to remedy your problem.

1.) Create a mesh of the same shape as your Label and apply it to the bottle with the Decal attached. This mesh can easily be created using the actual decal as a rotoscope template. The mesh doesn't have to be dense, just enough to conform to the radius of the bottle.

2.) Create a Transparency map and apply it to your bottle (my technique of choice). Texture mapping is very important to 3D modeling and animation yet often understated. For creating a beer bottle, there are easily 4 maps that you could create, all the same dimensions and all applied exactly in the same position. First there would be a color map. This will contain the color of the bottle (and any variations that may be in it) and the Label and possibly the neck wrapper. Next you could have a bump map that will slightly offset the label and raise any text that may be pressed on the glass (12 fl. oz.). You will also need a Transparency map to determine which parts of the bottle are transparent and which parts are opaque. You can also slight variations on the transparency to make some parts of the glass appear to be thicker than others. Another map I'd use is a Specular map (highlight map). The glass part of the bottle will probably have a much higher specular intensity than the paper label and you would use this map to determine that.

TommyDAQ

If you have the Beta, the Decals are now containers for one or more Images. If you click on one of those Images (by opening the "Images" folder in the PWS) your Property dialog will list (among other items) a section labelled "Tile" (or something like that) with an "X" field and a "Y" field. These are the number of times the texture will be repeated in a single stamping of the decal.

You should decal the entire area over which you want the tiles to be spread. You should probably flatten your model, although it will depend on what type of visual effect you want.

-- Tony

The rest seems to be pretty well covered, one snippet that may be useful to you is that the texture will be tiled within the boundaries of your stamp, so stamp the whole area you want to be tiled, then tweak the repetitions for the best looking size.

Matt Andersen

In 6.10, when you have added an Image to use as a decal, then added the Image to a decal container, click on the Image Name in the PWS, and on the Properties Panel is an X and Y value for tiling.

Jeff, Hash, Inc.

I don't know how many of us have been begging for this feature, but it's in 6.1 and I haven't seen anybody say anything about it! We now have the ability to place multiple decals of different types (bump, spec, etc.) on the same image stamp. No more writing the positions down and restamping! Just stamp a decal like usual. Now right click (PC) on the image header under the decal. Select import decal. A list of all decals in the PWS header will pop up. If it's not there, load the image in PWS first. Then select an image, and it will be placed at the same stamp as the previous one. Now it's easy to do a full color, bump, spec, etc. , or do multiple color decals on the same spot! You can even delete all decals from the image list, but the stamp will still be there. Perfect for decal replacement.

Nate

Decals are containers now. You can make a new container, add as many decals of whatever type you'd like into the containter, and when you stamp the container, all of the decals go to the *exact* same place on the model. You can also pick which of the images in the container you see when decaling....

Jeff, Hash, Inc.

Yes. Color, bump, transparency, reflectivity, ambiance, specular intensity, and specular color can all be saved in one texture.

Jeff, Hash, Inc.

Decals are automatically cached in v6.10b10

Jeff, Hash, Inc.

Nope. It checks the filedates. If there's a newer one, it loads it. If not, it uses the one in memory.

Jeff, Hash, Inc.

Nope, Only the geometry

First of all I like to thank the guys that have responded to my queries. I have experimented further since then. These are what I have found out about Flatten and Decaling.

• First of all the example in the manual is not that clear because a cylinder was being used. It should have been a sphere. Half a sphere should be ideal. A sphere is more related to human head rather than a cylinder.
• Half of the sphere should be selected on the longitude so that both the poles are included and hide the other half.
• Hide also the half of the the ring on the poles.
• On the Action window create a New Action. Select Muscle mode. Select the sphere.
• Rotate Z axis pointing the poles.
• Rotate Y axis pointing outward of the sphere. If Y axis is pointed inside, the Model won't be Flatten properly and you will get mirrored Decal or distorted Decal.
• The X axis would be parallel to the latitude.
• Move the axis to the edge of the sphere, parallel to the outer longitude.
• Select Flatten and apply the Decal.

- Johari

It's there. It is what makes up the whole back of the flattened segment.

That is because your decal did get a plied to your 8th section and when you unflatten the segment, it gets squished together.

What you might want to do is identify the 8th cross section and when you want to flatten the segment, hide that section so your maps don't get all distorted onto it. Flatten is really good for painting faces. And when you use it for that you usually hide anything you don't want the decal to appear. (the back of the head) and then flatten the face. Hope this helps

Actually the best way to do it is to hide the back half of the sphere and flatten the front then apply the decal. Then hide the front and flatten the back. Make sure one decal is for the front half of the sphere and the other is for the back. That should get rid of the squishy 8th segment.

Tim

Either. They folks over there developed an ATX file allowing you to use their textures directly in our software.... the tradeoff is of course, a speed hit, but they look awesome.

Jeff, Hash, Inc.

Yes. That is both of 'em. They have a plugin for AM98, "darktree.atx", which allows you to bring them in as procedurals and then add them to AM98 materials, like crumple or ripple. One thing that is really cool about it is you can create seperate darktrees for all the mapping types; bump, reflectivity, specular color, etc. and add all of these types to the texture you add to your material, you can get some pretty amazing looking stuff with the fully defined materials. Also, of course you can render out your textures in darktree as a wide variety of image types, which you can then apply as decals.

Matt Andersen

Well, I don't have -experience-, but I've done a fairly thorough read-through of the .itx documentation. I can regurgitate (eeew!) a summary of my take on that. Bear in mind that the Hash folks are saying (and I believe them!) that .atx will, in fact, be a more powerful interface. As you'll see below, this is saying something.

As always, those with more knowledge are free to correct me. I offer my correct information and my possible misunderstandings (as is inevitable).

- L O N G R E S P O N S E -

.ITX FILES

A '.itx' file is actually a '.dll' file, renamed with a different extension. DLL (for non-programmers) means Dynamic Link Library, and is basically a chunk of code that, rather than being hard wired into a program, can be run ... well, dynamically. Plug-ins for Adobe Photoshop, for example, are probably some variation on this theme. Hash will, at various points during its render, make program calls out to the procedure that is the core of the new programmatic texture. Specifically, it will call the program for each time that (in ray-tracing parlance) a ray strikes a surface. Obviously, if you write a slow .itx texture, you will bring your system to a dead crawl.

Because a DLL is simply a chunk of code, there is no limit on what calculations you can do during the call. You could calculate PI to three hundred decimal places. You could make a call out to the internet to fetch data. You could format the system's hard drive. Oh yes, did I mention that you probably want to be mildly cautious about accepting these files from people you don't trust?

The only limitation, from the point of view of -effective- programming is as follows: You can only see the information that the main program (for .itx: Imagine) allows you to see about its internal state. Moreover, you can only pass back to Imagine the information that it allows you to pass back.

Here is where the major differences between .itx and .atx will probably show up. Imagine's texture format was accreted over years of new revisions, and it shows. There are loads of extraneous pieces of information that the programmer -doesn't- need (but has to deal with anyway) and there are some fairly glaring instances of missing items that the programmer -does- need for certain effects. Without a doubt, Hash's eventual format will fill many of these gaps.

WHAT .ITX GIVES YOU

That having been said, here's what I've seen of the .itx format, and what it gives you:

• The first major item that your program will handle is what is called a 'Hit Point'. From the ray-tracing point of view, this is basically a point where a Ray has intersected one of the patches that has your texture assigned. You will get the following information:
  • Global coordinates of the hit point
  • Texture-local coordinates of the hit point
  • Distance the ray has travelled already
  • Direction of the ray from the camera
• The other major item is called (by Imagine) a 'patch'. Lovely, eh? Anyway, what -they- mean by that is a basic understanding of the nature of the surface you are hitting, and how it would be rendered -without- your texture. You will get the following information:
• Color of the point you hit (derived from any decals: remember that a hit point is smaller than a pixel, so anti-aliasing is not an issue at this level. Your effects will alter the rendering at the 'ray' level, and that will be aggregated into an anti-aliased result).
• Brightness, Reflectivity, basically all of the attributes you can assign to a material.
• Surface Normal at the point of impact (a vector which tells you, relative to the incoming ray, what direction the surface is pointing in 3-D space).
• In the 'miscellaneous info' category, you are also passed a whole bunch of programmatic junk that Imagine insists that you deal with. Man, I hope that Hash bypasses some of that.

WHAT YOU CAN CONTROL

In terms of what you can return: basically you can change the values of -everything- that is in the 'patch' structure. So, you can change the color, reflectivity and so on to get the same effects as decals. You can change the Surface Normals to get the same effects as bump maps. The difference, of course, is that the values you are inserting are calculated programmatically, so they should (if you programmed right) be resolution-independent: i.e. you can zoom in as long as you like, and still see finer and finer detail (-unlike- a decal).

Things I can -immediately- imagine you could do with this, if you were willing to put in the effort:

• Volumetric effects such as, say, grass, fur, mist, etc. This is a hard one to solve generally, but special-purpose results could probably be done with some major effort.
• Programmatic colorings that depend highly on the angle of intersection: soap bubbles, oil sheen on water, possibly even glass caustics (depending on how the internal engine works).
• Automated cellular organic textures: scales, bricks, cobblestones, rust. Many of these will probably be given to you with the things that Hash is converting from the free Imagine textures.
• Depth dependent transparency effects: ghosting, non-volumetric mist, selective fog, underwater shading, etc., etc. Many of these are also featured in the ABTextures that Hash is converting to .atx

WHAT IMAGINE SEEMS TO HAVE TROUBLE WITH

Before you go running into the streets shouting "we're all-powerful!", you should know that the programming interface that Imagine provides does impose some serious limitations.

And yes, this is my transparently obvious set of recommended 'improvements' that the Hash folks might consider. I can't help it: when it comes to programming, I can't leave well enough alone. :-)

• You cannot see any information about nearby hit-points and suface geometry. This makes it much harder (impossible, actually, I think) to produce more 'artistic' shaders such as a 2-D cel renderer -directly-.
  One would -like- to say "render as a dark ink-line if the point is near an 'edge' where the depth of the scene suddenly drops off from one ray to the next (as where a figure is standing against a more distant background)".
  But, because you don't have access to that information, there's no -direct- way to do it. You can, however, run filters to turn out information like Surface Normals and Depth Buffer into the numeric fields of an RGB map, and then run programs to locate discontinuities in Post-Production (Premiere or Multiplane). It just means that you will have to render many times, then post-produce, rather than just see your output in Hash.
  
• You cannot effect the future path of the Ray. This means that really serious refractive geometry (like having lens' and optics that focus images in strange new ways) is still not possible. It also makes it harder to achieve really soft-edged specularity in a programmatic texture.
• You cannot tell where the Hit-point was in the 'original', pre-animated model. This means that, like Hash's other procedural textures, these are going to act strangely if you Muscle-motion the model, and they're going to mesh strangely at the joints between boned segments.
• You cannot directly interface with more than one image map. You can read the hit-value of -one- by applying it to an object and then reading in the color of the Hit Point. However, it seems like it should be possible for the Hash program to cache off as many as the filter needs, and then calculate the intersections, without resorting to such kludges as placing the texture in the Reflectivity channel. This would be very useful for getting direct positioning control over programs made with (somewhat unpredictable) cellular automata. You could draw, for instance, a 'food map' which indicated where the automata should be denser, and where they should be more sparse (assuming you also coded the texture to read this map). Obviously, this would be even cooler with an animated food map.
• You get no information about the Bone that the surface you have just hit is attached to. This makes it harder to simulate Environment Mapping as one section of a programmatic texture (since you cannot tell what part of the bone-local environment the point is facing).

So, that's my little spiel about what I learned reading the .itx manual. I hope this helps, and I'll be glad to answer any specific questions (although my answer may well be "I don't know").

-- Tony

Try this before doing anything else:

1. read in the Hash manual about decals. ('help' in the program, starting on page 83 of the V4 printed manual, or page 159 of the V5 manual).
2. Go to 3D ARK at http://scozone.com/3dark/hashfaqs/index.html and check out Glenn Wilton's TWO decal tutorials.
3. Try zooming in on the model when you take the screen capture if it skips patches.
4. It's hard to tell from the way you wrote the post, but you do realize that to decal the entire surface of the sphere with one decal means you'd have to pull the patches until all of the unhidden sphere can be seen from one view. The top and bottom of the sphere are not closed off so you could distort it into a flattened ring to decal it. See the 2nd decal tutorial to see how to "unflatten it". Otherwise you will only decal the patches that are first "seen" by the program. Decals are "projected on" like projection maps, but unlike projection maps they stick to the surface of the mesh.

Hope this helps. I am by NO MEANS a master 'decaler' yet myself!

Glen Crowell

With v5 you can decal in a pose, so you don't have to mess with copying/renaming .sur files or whatever anymore. Just flatten yr model in a pose, decal, a viola, it works.

Ken Morton

When in modelling mode, create a pose, but don't use the "flatten" command. Just hide the control points you don't want, and manually flatten the rest.

Then make your decal accordingly, and apply to this pose.

Mike Caputo

Okay here it is:

First and foremost read this entire thing before you attempt anything I say in hear, so you can understand it. I stress that you should have a Wacom table to help you out, because you’re gonna need to draw a lot of lines. Also, you need to know your way around Photoshop. So here goes. First model your model. Get a screen capture of it from the side or whatever. Import that screen capture in Photoshop. what I would do next is bump up the image size to 144 dpi or whatever you want. I choose 144 cause it's an old habbit. Anyways, copy that layer. The screen shot layer is going to be on top. Then add another layer in between the bottom and the top screen shot. I also clear the bottom layer so that it’s one solid color. Now you got your layer in the middle, you can start the coloring. Coloring doesn't have to be all that detailed. You just want the basic underlying color of your character’s skin. Once satisfied, add another layer. This one will be for the skin cracks.

Here's where the Wacom will come in handy. For a dino texture, I whipped out the Cinefex issue of Lost World and also that issue of Entertainment Weekly of Lost World. They both had great pictures of the dinosuars. Simply just draw the cracks in like you see it in the magazine pictures. Don't worry about making it awesome now, we'll fix it later. For now, just draw lines. Try to vary the width of the lines also to get bigger or smaller cracks. Also, I suggest using the soft edged brushes to get a less harsh crack. And use black or a dark shade of gray! Once you get your cracks just the way you like, it probably looks like chicken scratch from any angle you look at it. That's good. Oh yeah, don't for get to save your Photoshop file!! So save it. Now hide all the layers, except for the chicken scratch layer and the bottom layer. If your bottom layer isn't white, make it white. DON'T save this file over your Photoshop file. For safety, save it as a different file, perferably something that will describe it as a bump map. Not flatten it so all you have is a greyscale image. Now the photo processing comes in. Guassian blur the sucker, but not too much that you loose the definition of the lines, but enough to make it softer. I say about 1.5-2.5. Now you got a blurry bump map. Here’s a cool trick. To vary the widths of the map so that the cracks are thicker in the corners and thinner in the middle, bring up the “adjust levels” menu or just crtl-L. Now, bring the white level up until your map looks like an organic picture of your skin cracks. You know, like the top of your hand. Now, bring the black level up a little so you get a good range of black and white. Not too much or you'll lose the soft edges of your cracks. Now save it!

Not using the same map, save it as a specular map. Gaussian blur it a little more so the cracks take up some more room and darken it so it is blacker. This will make it so there will be no specular highlights in the crack crevices, and gradually let the highlights appear where no cracks are. you know.

Okay, so open up your original photoshop file again, the one with the coloring. Now opn a new layer. Have your bump map open and copy it. Click on the layer, try to do it by right clicking on the layer in the layer pallete, so when you paste things, it'll fit perfectly. Now you got a layer that's just like you bump map. Hide everthing else except for your color layer, and that layer you just made. What we are doing now is adding shadows to the texture. Usually I would do it as a diffuse map, but for some reason, they don't work in V5 yet. So anyways, we gotta do it the hard way. Now set that layer as multiply so the color from the bottom shows through. Now set the opacity of that layer to something like 35% so it's not too dark. Okay, save you file.

I like to work with alpha channels, so go to your color map and select the floating group. Now flatten you picture with the selection still there and save your selection. Now save that file AS a targa file and 32 bit. Now you have a color map with crack shadows.

Whew! In V5, apply your color map first. The there's a cool then where you can numerically place maps. Write down those numbers that it gave you and when you apply you bump and spec maps, enter those numbers so they are exactly on top of each other. Also set the transpaerent colors to something like red or blue so the white doesn't disappear.

Done!

Jeffrey Lew

I just gave it a try in a sphere cut in half, I didn't even have to "flatten" it. Then I decaled it with a star/space image and moved the camera right up to the edge and rendered a few camera movements and it worked just fine. I don't know why you need a material if you just want a TGA starfield bacground.

Jeff

I believe that if you uncheck the Show Decals option in the Render Options panel that the AVA file will not be written every time you make a change to the model.

Jeff, Hash, Inc.

I dunno if you can decal in an action, but you can definitely decal in a pose.

You just put the decal on the flatted object. You don't need to unflatten, as long as you flatten in a pose. The original model will have the decal correctly applied.

Not that I know of, not in AM. You'll need to get the value elsewhere (like by using the eyedropper tool in Photoshop) and then manually enter it into the color dialog. BTW you can also just use an alpha channel in your map.

Ken Morton

Remember that the brightness of the texture will be effected by the underlying ambience of the material you have assigned to the points. Select all your points, make them a named group, and start fiddling with the Ambience setting in the 'Attributes' tab of the Property window.

-- Tony

This threw me off too, when they first added it. There is no "unflatten" command like in v4. In v5 you first create a new pose, hide what you are not decaling, then do the flattening procedure as outlined in the help file and position/apply the decal. It is not necessary to unflatten as you can either keep, clear or delete the pose when done as it does not permanently affect the model and the decal remains. Note: when you clear the pose it usually doesn't redraw correctly and requires you to close the pose and model window then reopen it.

Brian

You can do it.... The thing to remember is that decal order in the worksheet is important, they are applied as they are listed, top to bottom. So if you have:

...
eyebrows
facemap
...

it won't work, but

...
facemap
eyebrows
...

will. You can drag the decals into any order you like...

Ken Morton

I don't think it's a bug - I think it's a limitation. For patches to be legal, they rely on the patches next to them. When you flatten a model, the patches split along a seam so that one side has legal patches and the other doesn't. Even in the book, the example cylinder they show has 8 patches. When flattened, it's only 7.

ed Lynch

If you flatten in Action, select Action, Clear. If you flatten as a Pose, delete the Pose.

Personally, I prefer to do it in Action.

Jeff, Hash, Inc.

Nathan,

I don't know exactly what's happening, but you can have multiple stamps of one decal on a model.

1. Hide parts not to be decaled, flatten if neccessary.
2. Right click imported decal name and pick "position"
3. When it's sized and in place, right click and pick "apply"
4. Click outside of the decal area to turn off selection
5. Repeat steps to apply another stamp with same decal.

Just tested it again here in Intel v5.18 to be sure it's still working. I saved the model and then reopened to double check and the stamps are still there and it renders. (Multiple stamps of same decal on all sides of a cube)

Brian

Decal order is very important. Here is what I have discovered.

First, you can easily drag and drop to reorder decals. Jst carefully move it in between two decals, and the "universal NO" symbol will dissappear and a line will appear showing where it will drop.

As for order, here is what I think I know:

1. Specular first
2. Diffuse second
3. Ambient maybe - haven't played with it)
4. Color
5. Bump

This is specificly per application. If you have two stamps within a decal import, and they go to different areas of the model, you are better off importing the decal twice (hope that makes sense)

This decal ordering has worked for me. There may be an official HASH order, but I'm not aware of it. And YES the progressive render will show things that the final renderer will not, so usually if it renders in the progressive, but not in the final, it's just an ordering issue and can be fixed. I've never been able to NOT fix it with reordering.

Dale Carman

The example in the book say 90 degrees but shows -90 degree rotation of the pivot before flattening. However, on the example of the cylinder, it makes no difference which way the pivot is rotated.

Pivot positioning is also crucial to proper flattening, so make sure to check it from the side view as well.

Jeff, Hash, Inc.

Have you APPLIED the Decal ? (Right click / Command Click ) once its positioned

When you say no change, If you also change the decal AFTER its applied, You will need to turn off 'Cached' in the decal properties pannel. This will reload the decal.

Glenn Wilton

I think everybody getting confused ... I am :-)

If you want a hole in your object .. use Cookie Cut (grayscale map) ... This will remove any material settings aswell, the Manual shows a good example of the leaf.

Another way is to create a hole is with a transparency map (greyscale) that will make parts of your object transparest ... but note that specular hilights should still reflect off the surface of the area of the map thats transparent (like glass). The advantage of using a transparency map is that you can define Multiply or Addative transtarency.

(To stop specular hilights with transparency you can use specular maps etc .. but then why not use cookie? )

You really only need to use an Alpha channel in a Colour map. The alpha channel sets the amount of colour that shows through from the decal, Black .. No col, 50% gray ... a 50/50 blend of patch colour (underlying col) and Decal white Full decal colour. (I think I get the black/white arround the right way but you get the idea)

You can also use Alpha for bumps etc .. but its just as easy to leave a protion of the decal blank and the bump map will have no effect anyway.

Glenn Wilton

Displacement maps ARE in the program. Hash (Martin, Steve, Jeff) assured me that they worked and I would assure them that they didn't work.

Talk about mis-communication. They DO work. But the the numeric imput to get the same kind of displacement as V4 is very different.

In V4 you had to use WHOLE numbers (no "3.25", only "3" or "4") and each number represented an entire UNIT in Hash V4 (100 = 100 grid squares of displacement for white!)

With each build I'd eagerly test displacement maps without luck. Then I just went wild and typed in 50. VOILA! Displacement maps! I haven't begun to test them for the best range of settings, but it is obvious that V5 displacement maps work different than V4. I think that it is now a % like Bump maps.

Everyone: think narly terrains and mountains, think crocodile bumps!

(Remember to use soft gradual gradations from black to the lighter - raised - parts with the greyscale map so you don't "rip" your model. Keep displacement away from 3 point patches as well. If you animate and render you'll see why. I haven't tested it with HOOKS, yet. And remember that you need a lot of patch density to get the best effect. Each patch is subdivided 8 times, which equals 64 points of displacement per patch.)

Glen Crowell

Place your decal in the upper left corner of the wall object, select Apply. then take the little handle on the left side of your decal and drag it to the right past the original right edge of your decal. What you just did was flip your decal. Now Apply and then again drag left over right. the edge that doesn't move provides a nice seamless area. Of course you really have to pay attention to this when designing the texture or decal.

Another option, with less room for error, is to take a screen shot of the model, the open your texture in Photoshop, specify it as a texutre, then do a texture fill on the area that gets the map.

Quick, easy, and no mistakes.

Jeff, Hash, Inc.

I get lots of mail-order catalogs all the time for clothes (they come addressed to my wife). Some of them are very good in their print quality, and some favor head-on beauty shots of the clothes.

Scan 'em in at a high res and you've got a great starting point for clothing decals.

Hash's "poses" are great -- first import the tga file as a "rotoscope" and then flatten your object manually to match the decal. Then import the tga file as a "decal" and apply it. A perfect match.

You can save the pose in case you want to change anything later, or reapply something.

-mike

Try to build something simple a flat plane will do. In photoshop write something like your name. While it is still selected go to the selection menu and save selection. In the option window check save as channel #4. That's it you have an Alpha channel. Check in the channel palette it should black and white leave it like that. Now paint your letters as you wish and save. You can choose Pict or Targa. For me it is Targa and don't forget to check 32 Bit (RGB+Alpha =8+8+8+8 = 32).

Now lets go back to Hash. Right click your model in the project window and choose import Decal. In the construction palette you should see a preview of your image. Instead of color select cookie-cut. All the black area in your Alpha channel is transparent. Position and Apply the decal. Render and relax you now have your name floating in the window without any plane visible.

Frederic Drago

Hold the shift key while scaling.

-mike

Okay brother, put down the mouse cord, get off of the ledge, come on back in! Okay, here's what ya do... Create a new pose. When in the new pose, select muscle mode. Then select all control points. Now you can either use scale or rotate, I prefer rotate, but I don't think it matters. So let's say you hit rotate. You're going to want to rotate the axis, not the model, the model should stay stationary. To rotate the axis and not the model, you need to grab the inside handles as opposed to the outside handles that are used to manipulate your model. For your tree, you're probably going to want to have the blue axis(blue line) vertical, and the green axis(green line) pointing away from the back of the tree, in a side view. What's gonna happen is the blue axis will be used as a reference for the program to create a cylinder around it, then the program needs to cut the cylinder from top to bottom so that it can open it up and lay it flat including all of your model's control points with it. So it uses the green axis to decide where to slice the cylinder. If you've got the green where I had suggested, it will take it's imaginary scalpel and slice it at the back from top to bottom. Then when that's all done, go to the front view, right click on your model, and select flatten. That should give you the result you're looking for. It took me a while to figure out where to rotate the axis, try it a few different ways to see what kind of results you get. Hope this helps.

Sean

I always flatten my stuff by hand in a pose. It's much more accurate, I can leave room for where I want more detail in the decal, and it's just plain easier..

I just create a new pose and select the patches I wish to decal. Then I hide the rest. Then I name the name the group that's still visible and "mush" the control points until they're flat along one plane.

If you're looking at it from the front, areas that reach back into the z axis need to be spread out more along x and y, but you get the hang of it with a little practice.

-mike

There really isn't a great deal of reason to flatten 4 point cylinders, if you feel you must, hide one side go into a pose and do it by hand, once for the front and once for the back, personally I would forgo the pose muscular flattening, I doubt if distortion is going to be a big issue with 2 patches. If you use a higher density mesh, like maybe 8 point I think you'll find that flatten works like they say, keep in mind it doesn't just slice your model any old way, there has to be a row of points at x=0 for it to work properly. By the way in any future posts to the list, if you're interested in people actually understanding them,"the blue axis" is not a good defining statement, the letter x y or z, which means something, is in the lower left hand corner of your window, use that because MH3d comes with various color schemes and you can make your own so it's not blue on many people's machines. On to the mapping, select the front half of your model from the top 3 of 4 points name the resulting group front, do the same for the back, make a registration mark, this is basically 3 peaked 2 point splines that share a common point all snapped to grid, adjust this registration mark so it is within close proximity to your tree, but not inside any of its' patches. Change your zoom setting to some nice number that puts the trunk on screen large enough to paint on (or you may find it a better idea to split this into multiple maps vertically so you can get larger templates) and write this zoom number down so you don't forget, select the named front group and hide the rest then go to your screen capture utility, and capture the wireframe of the trunk and the registration mark, do this again for the back group. Go to your paint program, paint your tree in color make sure to go slightly outside the edges of the patches outlines or you won't get complete coverage, save the color maps, pull out the blue channels and turn them into bump maps. Go back to MH3d import your decals, select the "front" group again, check the zoom to be sure its the same as when you captured, position your map registration mark to registration mark then apply it same deal for the bump and the back side.

Matt Andersen

Only works on Planes, ie no Thickness. I had a very thin box, that I was trying to cookie-cut, sent it to Hash, didn't realize it was a cube. Took a couple of Mail exchanges before we figured it out. So Hair, which people on the list have mentioned being Cookie Cut, means that they're like partial spheres, like a football helmet, which is flattened and cookie-cut decalled, But there is only the 'skin', you can't have a thickness to the hairshape. (these 'hair' references are not something I've done myself, so correct me if I'm wrong.)

david

That's what I generally do. It works fairly well at double resolution, unless you think you're going to want to do some closeups. I think the highest I've ever pushed something was about 10x the screen-grab resolution, and that was for eyelashes (long and narrow application, so the resolution didn't translate to a file-size that would swamp the program). Just remember, if your total decals start approaching your system memory you are going to really croak on render times.

-- Tony

Yeah, you can definitely get a lot of impact in very little time with the atx textures. To access the itx textures, first you have to know where they are located on your machine, save them in a single folder. To actually get at them make a new material, expand it and click the attribute, select imagine from the properties window, say yes to any queries from the computer, the click the imagine file tab and point it to the folder where you have your itx textures saved and select one. To my knowledge the itx textures are currently uneditable, within this program, so what you get is what you get for now anyway.

Matt Andersen

The size of the model is the largest number of pixels that it will cover in any movie you make with it.

Generally, you should have a concept when building a model of whether it's going to be used for long-shots (very little detail necessary) or medium, or super-close-ups or what. As an example, on the Brawl pages (http://www.tmn.com/~albasch/Brawl/) (plug!, plug!), my Keitoshi character was built for medium-close action in an NTSC resolution setup. So, she looks OK at 400x200 (the large shot on her particular page) but looks much better at 200x100 (the icon picture on the main page). 'Course, this has nothing to do with the decals, but rather is an issue of spline density. But the same principles do apply to decals.

-- Tony

Since I've had quite a few requests for information on my water technique, I figured I'd just as well send this to everybody :)

Really, I don't do anything complicated, I just use a bump map. I gank the grid primitive off the v5 CD, slap an ultra hi-res bump map on it at 50% intensity, and then play with the raytrace settings until it looks good. In "Monolith" ( http://www.reno.quik.com/bprince/Monolith.jpg ) the water was set to a very dark green color, with about 90% reflectivity and 70% transparency. I usually use a touch of falloff too. Much of the realism of the water comes from the coral texture on the plane beneath it.

The only formula for doing photoreal surfaces is patience. I spent about 4 hours just lighting that image, and just as much time messing with my materials and decals.

An excellent source of natural-looking decals is the the cloud shader in Corel Texture. Its a little app that comes with the Draw suite, and its really an indispensible tool for any kind of work with 2D procedurals.

Brian "balistic" Prince

If you do manage to find some magical technique for creating flattened image maps, PLEASE post it to this list. However, I have been looking for just such a technique since I started working with 3D models and have come up completely empty. I think that you will find that there are very few tutorials or books that even cover decaling and those that do, do not cover it to any great extent.

Okay, with that, let me offer something constructive. Using Hash, I think you have two basic options for applying detailed image maps to complex curved surfaces. To illustrate these two techniques, let's assume we are attempting to map a face decal onto a modeled head. Next, let's assume that we are dealing with just the front half of the head - half of the sphere.

Techinque 1: DO NOT flatten the model yet. Apply the unaltered face image map to the head model. (I am assuming here that the image map matches the proportions of the model) If you were to render the head from the side at this point, you would see all of the shearing of your image map as it was stretched back over the models contours. Now go ahead and flatten your model. Save a rendering of this flattened image out to disk. What you now have is a template to which you can apply changes to from an image editing program and reconstruct all of the stretched areas. The old image map can then be removed from the model and the new flattened image map can be applied to the flattened model.

Problems: If you are looking to create a photo-realistic image, it is difficult to work with a distorted image map.

Technique 2: Do not attempt to create ONE decal for models with surfaces as complex as a face. Instead, flatten your image, (again, let's assume we are working on a model of a head) then save this flattened image out either as a wireframe or solid image, I prefer the wireframe since it gives you a general idea of where key areas are located. Next, apply a skin texture. This should be a fairly simple image map. Do not attempt to add any detailed or "position critical" features to this map. Keep it simple. Save this map and apply it to your model. To add all of the details such as eyebrows, lips, or wrinkles, apply individual decals on top of the skin decal. They can be merged seamlessly by using a feathered alpha channel.

Problems: This takes a lot of time and patience.

Solution: Use both techniques when creating decals.

IMPORTANT: As a side note, make certain that when you flatten a model it IS FLAT! If you use Hash's flatten feature all of your points might not be on the same plane. After using the built in flatten tool, scale the points to zero in the appropriate coordinates. This is important. Sometimes, when a decal is applied to a series of patches that are not on the same plane, the decal will drop off of certain patches in its unflattened form.

Sorry I can't be of any more help than this. I also hope that my text only explanation isn't too confusing.

Jason Tartaglia

AVA files are only created when Show Decals is turned on (Tools/Options/Rendering), and you have a map on your model, then you turn on shaded mode for bones......

Jeff, Hash, Inc.