Skaven252

Re: strange black squares Here. I was running the DX version in 1600x1200 with dual monitors (so my desktop is 3200x1200). ATI Radeon HD 2900 XT with 512Mb onboard memory. The screenshot was originally 1600x1200, but I scaled it down to 800x600 to save my quota; anyway, you can clearly see the black rectangles there.

My experiences with Alpha v62 (I ran the DX version): - I get strange black squares on the upper and lower right corners of screen, covering the menus. The menus work, but it's impossible to see what you're clicking. - First impression of pixel painting: AWESOME! Very fast indeed, and no problems with duplicate / overlapping UVs! (at least not with the quick tests I made). But as PhilNolan3D said, there are some problems with smoothing groups - When I tried changing the UV set with the Quick UV tool, the painting I had done previously was not transferred to the new UV set. (I didn't expect it to work quite yet, but gave it a shot anwyay) - After reapplying new UVs, painting didn't work at all any more, the brush just had no effect. I guess this is just a bug / unimplemented feature thing, related to above I didn't try it with more than 2 models so far, but it's looking very promising.

I'm not sure what you mean. ZBrush has "Cavity Masking", but it only works on dense high poly geometry - and only at polygon level - as it does it by calculating the convexity/concavity from the polygon curvature. I got the impression 3DC has a novel approach to this problem, that works at a per pixel level?

Getting excited... wow.... Now this is an interesting issue. How did you manage to do it with DP / low poly? I've been under the impression you cannot calculate convexity / concavity of corners, nooks and crannies unless you heavily subdivide the geometry. How did you work around this? Anyway, if cavity painting works on DP, it will be heaven sent for making worn corners on angular metal objects, like crates and guns and stuff. It's going to rule, and be light years ahead of any competition.

I'm a bit late to the party, but I wanted to say I'm really glad to hear this! Yay!

Oh, okay. I'll have to google what that means. Hmm, yeah, this explains it pretty well: http://en.wikipedia.org/wiki/Octree

The Filled Curve tool is awesome! Perfect for creating large, flat, curved objects like ears, wings, flags, et al. Wooohoo! \o/ I noticed that it's very easy to make holes into the voxels with the Smooth tool, if you create a thin Filled Curve surface. Since there are separate Surface and Volume tools, could there be a Surface Smooth tool what would allow you to smooth the voxel surface whlist avoid making holes? Or is there one already? About the manipulators: I think color indication on mouse float would already help immensely, without need for big modifications. Let me explain: currently the manipulator's elements (the Rotate circles, the Move arrows, the Scale cubes) only turn yellow when you click on them. So you're not always sure what you're pointing at before you click. Changing their color already on mouse over, ie already when the cursor is over them without requiring the user to click, would make the manipulators more intuitive. Then you would immediately see what part of the manipulator you're going to access.

This is just a thought / question that occurred me when fiddling around with voxels. How are the voxel volumes handled in memory? I would reckon that if there are large areas of solid fill or empty space, maybe they could be packed somehow to save memory. Also, if there are very small details in one part of the model, and only smooth undetailed surface in another part, could the voxel resolution change dynamically, so that the densely detailed area would have more resolution than the smooth area (kind of like "smart voxel subdivision" if you catch my drift)? Would the constant packing/unpacking of the data use a lot more CPU than just processing an unpacked n x n x n voxel matrix? I found a somewhat related paper on the subject, but I'm not sure how helpful it is (I think it has more to do with noise reduction than dynamic resolution): http://cds.ismrm.org/ismrm-2004/Files/002214.pdf

Add to the above: Being able to paint color, displacement and specularity, and preview it all in realtime. This is not possible in ZBrush, as it doesn't support texture channels the way 3D-Coat does. The Quick UV tool is pretty handy too. Plus, you can change the UV layout non-destructively as it's projected from the micro-polygons on the fly.

Sounds very interesting indeed, thanks for the tip! Too bad the bandwidth limit has been exceeded, so I couldn't view the videos.

^ Whereas InkScape is indeed an excellent free open source vector drawing program, it's not an API or library for programming, which is what Andrew was looking for.

Good thread and good points Jokermax, I agree 100%. I even thought about starting a thread myself with exactly the same title.

I've been wondering what was "wrong" with the Clone tool until I tried using it with a very obvious test case. I just took a cube and painted some material on its side: I then rotated the view, and tried cloning the pattern from one side of the cube to the other: But as you can see, it looks like the cloning (Translation) actually happens in screen space. The results get distorted by perspective, and so does depth/displacement. I suppose some people may find this useful, but it's not quite what I expected. The other modes just rotate around the source axis, or require Symmetry which is not always available, and is also restrictive: you can't Clone skin from the arm to the leg via Symmetry. What I would like to do, is to use the Clone tool to copy surface features (color, specularity, displacement) from one part of the model to another, without getting them distorted by view direction, and without being bound to symmetry. It actually does work just like this in the 2D view. But alas, in 2D view I can't paint across UV seams, it will cause discontinuities. Also, the UV islands have different rotations, but the Pen Rotation does not affect the cloning. So, the requested "3D" cloning tool's Source point would constantly align itself to the surface normal just like the paintbrush does, and travel along the surface of the model, unaffected by screen space. The copied features would be aligned to the Destination surface normal and projected there. In other words, it would be "Cloning in 3D view, that works like it does in the 2D view, but allows cloning over UV seams.... and also copies Specularity by the way". I imagine there may be some complications with more complex 3D models. For example, if you try to clone the skin color on a thin finger across a wide flat area (like the character's back), the Source point would orbit around the finger multiple times during one long stroke, creating a repeating pattern. However, I think most people would use Clone to copy from one relatively flat area to another, and understand where to use the tool and where not. Also, Z rotation may get tricky (example: fur pointing downwards at Source gets cloned as pointing upwards at Destination), but if there was a faint preview of the Clone Source on the brush circle (similar to the Material overlay but only visible within the brush circle, just like when painting a Cube mapped Material), and the Destination could be rotated by Z, it would solve this issue. This kind of adjustable rotation would also be useful for cloning in the 2D view, because the UV islands tend to be rotated around. Now... how would you make this work with Voxels? Maybe clone a spherical 3D volume (aligned), rather than a flat displacement?

Big thumbs up to this idea! It would also be nice if 3D-Coat could read the clipboard contents into any of those channels. Then it would be very easy to quickly pre-process (crop, high pass, what not) the bitmaps in Photoshop (et al) and toss them into 3D-Coat for use without having to save intermediate files.

Excellent! Just downloaded, installed and tried it, and can confirm it's now working right.

Here is the topic where this was polled and discussed before it was made into a feature request. In that thread I suggested a simpler way to implement this that will not require an UI overhaul (as in, Luminosity would be an additional channel, requiring new buttons and sliders on the UI, and adding a Luminosity channel in all Layers). The other way to do it could be to give one of the texture Layers a new "Luminous" blend mode. Then anything brighter than black painted on the Luminous layer would glow in the dark. Implementing the feature this way would require much less changes in the UI, and - while I'm not a programmer, I suspect it would mean less overhead in resource usage as well, as not every layer needs to have a Luminous channel reserved.

My turn! I used the creature mesh to practice and learn using 3D-Coat. Comments on what techniques I used: - The blue skin is the default "elephant skin" material painted thru the material mask. Tried some other materials on the back too. - I used Retopo / Use Current to add an extra edge loop for the eye. The texture from the old model was projected to the new retopoed one. Handy! - The rings around the trunk were painted by using the "Draw with splines" tool. It's easy to draw the spline around cylindrical shapes like these. - I used Cavity / Height masking in various places to put color / specularity in heights and cavities - I Froze the skin on the feet to draw the "trouser leg" edge. - The pink scale pattern on the creature's belly was a Hexagon volumetric fill, applied by freezing the belly, then inverting the freeze pattern to freeze everything but the belly - You see some inadvertent flaws here and there. They were caused by the Clone Tool menu click bug, and were not undoable. - I could have broken the symmetry on the creature's back by painting over it, but I didn't... All in all, it was a great learning experience.

I just tried XNormal. It requires four photos (illuminated from left, right, above and below) to make normal maps from photos. Makes sense, but in practice I suppose it's quite hard to come up with such photos, especially if it's a large stone wall or such. Not everyone can carry a directional spotlight with a rig with them when taking texture photos.

Hold down CTRL and click on multiple pens, they will all be selected and switched between randomly. Was this what you're looking for, or are you talking about something more complex?

In other programs this would be the case. But for 3D-Coat, it would be most useful if it was available as a Fill modulator. That is, you could choose a DarkTree texture as one of the fill modulation options. Being able to rotate and scale the built-in fill procedurals in 3D-Coat more freely would also be a very nice feature. Currently they can't be rotated (AFAIK), and Anisotropy only scales the procedural by one axis, in one direction.

Hmm, I wonder if that bug is akin to what I posted here? (maybe I shouldn't have made it into a separate topic?) There seems to be some artifacts introduced when voxels are projected to height maps. Something tells me it happens where the retopo polygons intersect with the voxels... but not sure if that's the case.

Ahhh, now it works. Thank you!

Is there any way to access or edit Smoothing Groups within 3D-Coat? They don't seem to work if I import a low-poly model that has them. There's the "Ignore" and "Auto" option, but they don't seem to work or look right in low poly. And is there any way to change them afterwards, if the model has been, for example, vox sculpted and topologized entirely within 3D-Coat? How are smoothing groups handled within 3D-Coat?

I would second that for the Render panel. It would be handy to be able to render clear images of your normal textured 3D models as well. Currently you have to either take a screenshot or import the model and its textures to an external application to render (unless I've missed something?).

Yeah.. forget what I said earlier, I actually tried it in practice this time. I had an existing low poly (game optimized) model with an existing UV layout with overlaps. I tried importing it in a multitude of ways. First it ended up looking like melted ice cream, just like Phil's screenshot in the previous page. But I finally managed to get it just about right. But even at the lowest carcass resolution setting it still subdivides the model once, and the UVs get stretched. Here's a little shot of the model, filled with brown noise, so you can see the stretched texels: There is a plus side though: cavity masking works even in low poly! Hooray! I used it to Desaturate and Specularize the sharp corners of the object (see above). I love being able to do this so quickly, making a metal object look worn used to take hours of tedious drawing on isolated UV islands. Another cool feature is 3D-Coat's automatic UV packing. I decided to "Keep Clusters", but let 3D-Coat automatically assign the object unique (non-overlapping) UVs. They were packed very efficiently. However, when I tried using the quick UV tool to re-unwrap the object, the clusters got rotated and scattered quite inefficiently. Anyway, the model above is poorly suited for painting. The panels on the front are not Booleaned into the surface, they are actually separate boxes with no backface (to optimize the polycount). Thus, there's some face/fill overlap which will probably cause problems with direct painting. Anyway, I'm really looking forward to the direct UV painting feature. Here's hoping that cavity masking will work with it the way it does now. Because I got the impression somewhere that it requires tesselation.. so how exactly does 3D-Coat do it? There are other problems to solve too, still. For example, if the UV clusters are at different scales (for example, I tend to scale the object's bottom face UVs smaller to allocate them less texel space, just for optimization) and you draw a paint stroke across the UV seam between them. How is the scale difference handled? I've seen many 3D paint programs struggle with this. But I'm a firm believer in Andrew & co's amazing talent.