Moderator note:if you want a higher chance for being front-paged, make a topic and include the link in your reply.
I've seen a lot of cool threads around here recently, from LitSpheres to Snow blending shaders etc...
It's a pity we wait to have proper finalized stuff to show it...
So what about a technical wip thread?
I also stumble upon tons of cools papers but i never share with anyone...
There was a similar one at tech-artists.org but there's just too few users.. Is polycount big enough? We'll see
I've been lazy for almost a year, but we have to start with something right? SO, I dug up some stuff
Spherical lights:
Some might remember these endless talks about how point light have no accurate specular blablabla, so I came up with this simple model where lights have a defined radius. (
http://www.mentalwarp.com/~moob/show/polycount/sphereLights.jpg)
I recently updated this shader to generate a procedural floor reflection and to take the new model into account into the diffuse contribution, and not only specular.
Yes it is expensive as hell, but i was bored at Xmas, forgive me: lights are calculated in a loop in the pixel shader, and the process is done a second time just for the floor (even tough in a simpler manner: no normal, no spec) (yes i cheated there is precomputed occlusion/shadows)
Multilamp BRDF shader, yay!:
I've been working with unity on iPhone, which only supports 'Sphere Map' instead of regular cube. So i had to write a shader to transform cubes into this funky oldschool format:
.
Some of my recent work at CCP got released not so long ago:
[ame]
- procedural moons (i did selfshadowing craters at first, but it's not very old cards friendly...)
- animated gaz giants with moving clouds
- planetary rings with procedural cast shadow
Let me see your stuff now
Replies
I've got a couple of WIP MEL scripts and Maxscripts, will see if I can grab any useful screenshots tomorrow and maybe polish one or two ready for release.
A maxscript LWO importer, a Maya "OBJ bits box" and some random UV tools I think...
I tried the LWO importer for MEL and it never worked, but maybe now since I have a better understanding, I can take it on again.
[ame]http://www.youtube.com/watch?v=tEWqvrXaJE0[/ame]
I've been working on a lot of windows tools lately. The most promising is a sort of pie-launcher, very similar to the one in Maya, or Modo. Or if you have a Wacom Intuos4, it's a bit like that one.
It's used for launching files, apps, etc.
I got busy with some work crunch so I had to put it aside, but I'm going to pick it up again soon. I'm just about ready to start having people test it.
if its not already implemented could you do a snap to pixel center function for finished layout,
(selects all vertexes and snaps it to a pixelcenter given the dimension of the map)
(input height with button output -> snapped)
the lua file describes the content (widgets...), and I can have some fun with automatic layout management.
Personally I'm working on a 2d unity3d clone for XNA. It's not really pimpready though :P, maybe in a few months.
Anyone have any cool to share regarding WM2?
The coloring of the terrain is just for preview purposes, to see how my masks will work.
I use one monitor at home, at 1920x1080, I don't work with the nodes that small.
But yeah, Windows 7 default or whatever. Can't be arsed to change what I don't find broken.
I do have a question though, would it be possible to have it export out at preset sizes already? I noticed it was at a lower rez (800X800) when you brought it back into photoshop,
Is it possible to keep it for higher end detail when copying over so if your texturing lets say a 2048 based model etc It gets good pixel depth? Or are those all in the works right now?
With that again I could compute a scale ratio (by comparing visible texture space and screen view space) and scale the perspective output by that. I think I will add that, that might be a nice way to go to determine the optional size for painting on based on the texture size.
I will add some checkbox for switching to either viewport renderer or CPU renderer.
yeah if they stood at set res's that you could calculate based on 128X128, 256 X256 Etc that would work great...though would it work with odd set dimensions? Like 128X256?
This is a great thread idea and I've enjoyed reading it so far. Hopefully it continues to grow.
The new eve stuff is lovely! Round of applause for you.
I have and always will be working on my coolRIG system.
[ame]http://www.youtube.com/watch?v=bzXsGeBjv50[/ame]
Flixel Lip-Sync Example
Altough the exported table should also be usable in other real time 3D apps/games.
Most adjustment layers are currently supported including different layers stacked using various blend modes.
http://boards.polycount.net/showpost.php?p=1056535&postcount=665
Several things have been tweaked for the script:
- non visible faces (backface culling) will not be rendered in the texture map and instead have a magenta color (255 0 255) that you need to filter out in Photoshop (can't copy transparent bitmaps to the clipboard).
- only after successful operations the radio button for the other operation is switched. For example if you forget to copy the painting back to the clipboard it will give you a warning message.
- a camera object is created for security reasons. Once in the scene it will be reused and transformed each time you take another viewport snapshot. That way should you by accident move the view it still refers to the former view because it uses always to that camera when rendering to a texture map (used for the camera map material).
all in all a nice script where you don't have to worry to much.This weekend I will try to capture some workflow video of a nicer example while trying to extend a scene with custom paintings for different views. It should be the perfect way for me to test it.
Damn awesome, trying out the camera painting right now!
Looks like a gras/fur surface shader where the direction is the derivative of an accumulation buffer
@Richard Kain, that's pretty awesome too, illusion is there
I just did a quick test of your camera mapping script:
It's working really smooth, very easy to use, thanks a lot man! There's a lot of potential with this , I feel like I'm making 3d concepts now!
Do you have any plans for making it so textures can be baked down with alpha? So It would be possible to work in layers instead, like you only bake out what you painted in the latest step.
Another Thing that I'm missing right now is some padding on the baked out texture, I usually bake down with 32 pixels padding or so to avoid any ugly visible mipping seams.
*EDIT* nevermind, I see there actually is some padding, i just baked out on such a huge texture I didn't even notice it . Is there a way to set it to more padding?
*EDIT2*: Another thing that would be nice if it's possible to specify the size of the view to be rendered out, like with my example it would be useful if I could render it out larger than my screen size, to preserve as much detail as possible when I'm painting over it in photoshop.
can be defined by changing the input value next to the tiny "G" (for grabbing texture size info). Its the padding size value, change that value to lets say 32. The other 2 input textfields are the width and height of the RTT texture size.
I still have to evaluate some options for that, because I don't know if I can setup a custom viewport or preview rendering size. Otherwise I'd have to go for rendering using the software renderer which might actually be more precise (filtering,...).
There are many things that could be improved and added like :
Some bigger issue I noticed myself was the lack of proper filtering especially if you have lots of small uv- shells. So I worked on this light match to view script that will create a shadowmap that I can use to filter out invisible elements from that view in the rendered RTT texture map.
Before it was only removing backface culled faces but not areas outside the camera or hidden by other objects in front. Using this technique there should be a very nice masked area instead of wrapping and looping stuff around like now.
Download it Here - simply extract to photoshop directory.
doesnt pressing alt toggle between dodge and burn by default?
do ppl use dodge and burn much? i rember i used it back when i first started texturing but after i learned a bit more i've never touched it cept for very specific things
Most people don't use this, but I use it extensively for metal.
It helps me fake reflection when only specular is available, and unfortunately I can't find an alternative to this time consuming method. (so this script helps me a bit)
AS for Alt, it only switched the tool but not the settings, so if you're like me and set dodge to highlights alt would ignore that.
This one toggles between standard and full screen (with menu bar), instead of cycling between all the modes. (I only use these two anyway)
Download qToggles_v0.2
I wanna see if I can somehow map these two scripts to normal keyboard keys like O and F.
It's just a pity that the two things I want to mess around with is for Max only (TexTools and coolRIG).
I currently have it setup with version control at work so if someone adds a button and checks it in the next time another user gets latest they will have the new button, and that user can hide the button with a right click menu if they dont like it. It also automatically generates macroscripts for each button so you can bind any button to a hotkey.
Heres an example code for a button.
You can also simply use the filein function in the button expression to trigger a more advanced action like this:
Will try asap! I am tired of the huge default max toolbar buttons...
so how does one configure the layout and the set of buttons, is it within a maxscript file or some use of ini or xml files? Resembles a bit what I tried to start some time ago with MZPrucksack
Render, how does your script behave so far? Last time I checked it out it looked a bit complex to setup. Honestly if Max allowed custom sized icons and expandable/collapsable toolbars it would be close to perfect ...
Also all the button icons are modified with an overlay shading image when the toolbar loads, and non-square images will be automatically rotated if the user has the toolbar set to be horizontal instead of vertical. There is also a library of common code for things like string parsing, bitmap manipulation, and array stuff.
So directX is winning again here, but not because it's awesome, just because of the lazy fuckers in the other team.. grmlml