First of all. Yes somebody is thinking of helping me with learning python. But can this even be done in python ?
This is now in html. And is very plain and clean.
http://aprilcolo.com/oh/tank/folwy/paint/paintdawn.html
What do you think ? 
that is a slow server
it am getting it at sub-dial-up speeds (I have dsl)
bah, now iâll ramble:
(Note: I havenât poked at the blender source much)
there are many challenges in coding a good 3d painting engine:
- How do you tell which face (or faces) the brush is painting on?
Well I guess that is the only thing that came to mind.
My solution to it would be to give each visible face itâs own color (limits painting to 16.7 million or 4 billion faces depending on depth) and correspond the painting cursor with that image, and the color in that image to a face. This image will not be visible to the user, there is no reason why it should be.
(yeah, blender did that)
The brush will have to warp as it moves along the texture for a number of reasons:
- The uv mapping may not correspond to the actual model. I.E a sqare face could have a trapezoid shape on the actual texture.
- the person painting will not always be painting said face straight on, but should probably have still a round brush (from their point of view)
- The faceâs texture and the uv texture may be rotated differently. If painting with a non-round (or an image) brush things will not work right.
and the user can be painting along the seam of two faces which are non-adjacent in the texture. It is probably necescary to consider which faces (plural) the user is painting on, which would require further refrencing of the face index image.
of course the face index image would have to be regenerated every time the model is moved. It doesnât make sense to be able to paint while the model is moving (well, not usually), so only when the view tumbling has stopped should we generate that image.
see, and what about quads, having the problem that you can see both sides of them. will the image the user sees have triangluated quads, and they paint on them (well, they are drawn triangluated now). I guess with any reasonable mesh this isnât an issue.
there a bunch of tricks that can make it fast, look at the gimp source for ideas.
well I did not go into great detail about vertex paint. But my notes for it had spoke of Vertex Paint as the possable answer. As vertex Paint can automaticly find the next vertex. And if you turn on faces while vertex painting it finds the next face. So their is something there.
And my server. My server is a pro server company with the home down type. I am on dial up and it is speedy for me. Even on window machines at school. So I have no clue why it is slow for you.
It is faster now (took about 2 seconds to load)
but 1.5k/sec is REALLY slow
now then, vertex paint is probably faster because the painting algorythms blender uses is probably slow.
I.E. Taking a texture, changing it (is it done on the cpu or video card) and then sending it back to the video card. Not a simple task, but one that can spread load better than the way blender does it now.
I kind of belive that is why blenderâs painting thing is so bad now, not finding the faces, but painting the texture and updating the view.
Ideally this would be done on the video card alone, but even stillâŚ
I guess I have to look at how it is done in source now.
Bah, well anyway blenderâs painting thing needs some new features
- colorchooser and swatches (needed by materials and vertex painting too)
- A way to paint more than one color channel in an inteligent manner
- a preview that is more accurate to the render, but possibly options to disable certain channels?
- speed and flexiblility
- more brushes, and options for them (animated and fractal?)
- more blending modes
- possibly layers for each texture?
Weâll see, but I have other things I should be working on I guess.
but painting is mondo fun ^v^ 
. It is akin to 3d modeling. You can see it and tinkerer with it.
I always download blender when I am at somebodys new computer to test and play in it . But I can never finish anything to texture phase because of blenders underdeveloped uv unwrap features.
And I always wanted blender to be an end all end all of apps that has everything needed for play and work.
Features like an IRC built in. Html/css mark up. Flash export with markup. Painting. Would be so nice to have in blender.
So far it is a 3d, Game, Text, Movie editing force.
^v^ stuff. I am willing to learn and help out anyway needed.
UV isnât all there is to texturing you knowâŚ
Martin
yes so very right. Talent and skillz are also needed. But at least a smoother and easier texture mapping method would help newbies. While giving pros a new tool to exploit and create maddning cool images that would not have been so cleanly done otherwise.
no no no, I meant to say that UV texturing isnât the only texturing option there is. Youâd be a fool if youâd UV map everything.
Martin
But for fine details it is the only way to go. I have yet to see a shader that will create a dragon tatoo on the model wherever I choose it to be .
Placing decals with object mapping and an Empty.
Given, for character animations, UV mapping is a sure bet if you donât want the texture to ripple with the deformation, but making the texturespace sticky works great too.
Martin
okay but what about exporting to another program. What then. I doubt sticky will be in the other programs. But uv textures will be.
And this would also be a good tool for texture animation. And 2d cartoon making.
But hearing from the irc and mailing lists. I cannot find any coders that can even figure out my simple idea. what more can I say.
I think a lot of people understand what you mean, its just projection painting. Like you can find in programs like DeepPaint, Bodypaint etc, its not new, the easy but is saying what it is you want, the hard part is figuring out a method of implementing it. The problem with basing it on vertices is it relies on the topology of the model to add detail, whereas a bitmap textures detail relies on pixel dimensions of an image map - two very different concepts.
I think the reason few people are following you, is that the proposal needs to be more in depth. You need to clearly state the objectives of your proposal, some links to examples & perhaps to some papers with research done into similar implimentations. Dont get hung up on the clone brush, thats just one tool among many that should be amongst a good texture painting implementation (thats details, not the main point behind your proposal). Dont get hung up by one specific instance of what you want to do & think more generally about what a useful tool could do. Dont go of on tangents about how fun or cool it could be, people should be able to work that out themselves from the information provided in the proposal - just telling somebody that something is cool doesnt convince them it is, you need to back it up with facts! 
_proposal - itâs O.K., but (of course, you donât have to agree with me) the most immediately useful feat would be to apply something like atlas unwrapping method (from lightwave), and to revert to cube mapping as known in 223 and leave painting to apps of choice of illuminated users - until drawing tools in Blender mature fully.
I ainât illuminated, I speak strictly from practical point of view.
But I am sure I wouldnât like to have irc and likes integrated in a production package, whether it is open sourced or not.
cannot find any coders that can even figure out my simple idea. what more can I say.
http://www.python.org/doc/current/download.html
http://www.brunningonline.net/simon/python/PQR.html
say yeah-yeah
As much as I want this major tool in blender, I just donât have the heart to keep fighting for this feature that we all needed along time ago. And it was almost created in the past when the first texture painting was created for blender. I was their asking for it then, but they never did.
So I guess that the only skill that could be added right now is a better uvwrap from python. But I hear that the new api is still missing some code libaries.
So that is that. Well hopefuly one day somebody else will want this feature. And they can start from where I failed. 
I myself have the amount needed for Zbrush now. So that will be the route I will choose for texture paint. But I sooo wanted blender to have it. As Zbrush is a one users thing. And nobody can share it.
Weirdhat has offered to help me learn Python. I am just waiting for him to respond. If you want to teach me a bit I am willing to learn. But me just staring at docs and pdfâs and books I will never learn it that way.
I guess I will make a proposal on âteach meâ ehh I will have to think of that a bit.
^~^
I donât think python is fast enough for this
uvmapping probably, but painting is another matter.
To complicate things more (and increase speed often) newer image editing programs store images as tiles of like 64 by 64 pixels (to change come 64 bit architectures) to make editing of the image faster, you arenât accessing ram all over the place (it is ârandom access memoryâ but apparently there is a penalty for treating it as such)
baa, shall change avitar now
baaaaaa (I love that show)
baaaaaaaaa I will just spend money for the real thing www.zbrush.com
Download the demo and you will see what I was hopeing to get a fraction of in blender
Youngbatcat,
In terms of what most other leading programs in the 3D industry are doing Blenders unwrap features are hardly âunderdevelopedâ.
Blender may be missing a few features but none are critical to the process. Yes, there are some really cool uvmapping tools out there, (like the ones found in Wings3D for example) but for the most part âprofessionalsâ are still using planar and cylindrical mappings which are then edited by hand. You need to learn the tools that are available to you.
here is a tutorial of how you might uvmap in Blender:
http://mywebpages.comcast.net/jmandmc/uvtut/uvtut.html
TorQ
Youngbatcat,
In terms of what most other leading programs in the 3D industry are doing Blenders unwrap features are hardly âunderdevelopedâ.
Blender may be missing a few features but none are critical to the process. Yes, there are some really cool uvmapping tools out there, (like the ones found in Wings3D for example) but for the most part âprofessionalsâ are still using planar and cylindrical mappings which are then edited by hand. You need to learn the tools that are available to you.
here is a tutorial of how you might uvmap in Blender:
http://mywebpages.comcast.net/jmandmc/uvtut/uvtut.html
TorQ[/quote]
well I sort of (just a little) beg to differ. True ALL major programs still just use basic unwrap tools, without plug-ins. But they tend to have more and better unwraps. Like atlas and True stretch flat cylinder.
Maybe if the old PolMap from Eeshlo or somebody, was recreated to work in 2.29c that could help. But even still the seam problem exists.
The power of the texture paint would be to âquicklyâ and accurately solve the UV seam problem.
Also . I must ask. Has anyone here downloaded a demo of any of these programs yet ?
Zbrush www.zbrush.com .
Textureweapons-Deep Paint. http://www.righthemisphere.com/products/dp3d/
BodyPaint http://www.maxon.net/pages/products/bodypaint3d/bp3d_r2/bp3d_2_e.html
IFX http://www.ifx.com/pages/amazon/overview_linux.shtml
Nendo and Mirai http://www.izware.com/mirai/product/2d_and_3d_paint.html
Teddy 3d http://www-ui.is.s.u-tokyo.ac.jp/~takeo/java/smoothteddy/index.html
You realy need to try them. That way you can understand why I am so head strong about this issue. It shows the power of what is coming in the next coming of 3D. And blender should jump on the band wagon.
At least try them for a minute. And if you are on dial-up (like meee) just let it downloaad in the background they are worth it.
Also that is my next reason for getting it to blender. Blender is such a nice small download and texture paint would be so nice in that download when I am out and about stucck at someone elses computer.
Hi,
I am unfamiliar with âatlas and true stretch flat cylinderâ but I do know a thing or ten about DeepPaint3D (I use it everyday at work). DeepPaint is not nearly the great tool you (and I) are imagining it to be. For one, DeepPaint still depends on your model having really good UVmapping. If you are thinking âhey but what about projection paint mode?â I say think again. Projection paint is pretty cool because you can paint resolution independantly. When it comes down to doing what it is meant to do, namely painting over seams, projection paint fails miserably and still is reliant on good UVing.
Teddy on the other hand (and I think Nendo works the same) is another story. Yeah, Teddy is cool because you can just start painting and it does its own UVmapping thing. The REAL problem with this is no one can go into their paint program and edit the map itself because the model is basically exploded into triangles. Why would you want to go into a 2d paint program anyway? Well, as you know in order to do away with tools like photoshop you have to have a pretty damn good 3d paint tool and none of the ones I have used come close to photoshopâs power.
Iâm going to look into Zbrush by the way. (EDIT: looked at ONE example of ONE of the auto UV tools and it basically blew up the mesh too)
I agree that Blenders 3D paint tool could be better but I think it really starts with just making the basic brush better, having a color picker in the 2d window, and making it so it doesnât smear when you paint across edges.
Better UV tools would be great but until someone includes the autoUV tools from Wings3D you can allways export to Wings and get really nice UVs and then export back to Blender. Hereâs a tutorial I did on the subject:
http://mywebpages.comcast.net/jmandmc/uvmap/uvmap.html
yes, I know, it requires some amount of effort on your part, but then again most good things do!
Take care,
TorQ