Jump to content
3DCoat Forums
Andrew Shpagin

3DCoat 4.9 BETA testing thread

Recommended Posts

Hi everyone !

It's been a while since I didn't report anything here, I'm having an issue in painting room around seams, as depicted in this video :




Here are two big problems :

1 - painting over seams is no longer reactive, since you often have to hide then show back again your active layer to diplay what has really been just painted
2 - some pixels remains definitely "unpaintable", which causes serious problems when you need to texture something...


I'm usually running the GL version, but DX fails the exact same way.
I'm also encountering this bug since several builds (at least since 4.9.55)
Here are my preferences : 


image.png.accffbc7d8ab5b9700eeced206fa062a.png
image.png.79be5a9c16d732030119d111788283ec.png
image.png.fd9423365b1b443111c60ac70c9cde4a.png
  • Like 2

Share this post


Link to post
Share on other sites

I wish you'd shown the uv-map, because unpaintable triangles are usually (always) because of unwrapping. Automap will do that often.

also it would be good to switch to flat shading (view menu or 2 on alphanumeric keyboard)  to see if the triangle is really not painted or if this is a shading artifact caused by either a flaw in vertex normals (maybe a quad triangulated the wrong way) or in the normal map (bad unwrapping).
 

Share this post


Link to post
Share on other sites

It's made using the default robot provided with 3DCoat (shoulder part) + it's already in flat shading view. What you see is only curvature(overlay)+AO(Multiply) over "Layer 11"

So, to reproduce :
- simply launch 3DCoat
- chose paint mode
- pick this little robot
- activate flat mode
- toggle seams display on
- create a new layer
- pick a colorful tone (red, blue...flashy helps better spot the bug)
- paint over seams (on my side it fails over the right shoulder)
- you might see a lot of fails
- hide layer
- show back layer > most "unpainted pixels are painted now (1st bug)
- some pixels may remain unpainted and these will stay unpaintable (2nd bug)

Appart from this, past versions of 3DCoat used to create proper pixels on seams correctly when painting, even with an automatic unwrap (which made it useful on rush parts of production when you can't spend time on UVs, quite rare, but happens).

I've got the bug on all my models since several versions, and I often do and control my UVs precisely to get the cleanest bakes possible (I don't like automatic stuff...). I've tested with 4.9.05, quite same issue but closer to what you've called "triangle not painted" in this one. (I'm downgrading to 4.8.44 to see if I got better results)

Edited by Vexod14

Share this post


Link to post
Share on other sites

I think best to show the problem on another model.
That one has a lot of problems. I think it would require to change the texture resolution to make it much higher to be able to paint on it.
 

there are sharp edges all over the place, pixels that can't be painted indeed, and the ambient occlusion layer has black spots on the outside surface,  etc... image.thumb.png.9eaded6034bfaf6c5aaba1a91189c4bb.png

Share this post


Link to post
Share on other sites

Just to be sure we're talking about the same unpaintable pixels : You see that little grey-dude on your screenshot ?
image.thumb.png.d608cb289aca362f40db5f2b3c566b6c.png
That's exactly this kind of issue I'm having on each models around seams, I use to split vtx normals over seams all the time that said (but I'm doing so since 2015 and I've only seen that unpaintable pixel bug quite recently).

By the way, on the 4.8.44 I don't have the 2nd bug (which is the most problematic), but first one was already there

Share this post


Link to post
Share on other sites
1 hour ago, Silas Merlin said:

I think I saw this mentioned somewhere but I can't find it with search :

When using auto pick, if the current selected layer happens to be ghosted, auto pick won't work.

Video is here =)

 

Share this post


Link to post
Share on other sites
21 hours ago, Vexod14 said:

Just to be sure we're talking about the same unpaintable pixels : You see that little grey-dude on your screenshot ?
image.thumb.png.d608cb289aca362f40db5f2b3c566b6c.png
That's exactly this kind of issue I'm having on each models around seams, I use to split vtx normals over seams all the time that said (but I'm doing so since 2015 and I've only seen that unpaintable pixel bug quite recently).

By the way, on the 4.8.44 I don't have the 2nd bug (which is the most problematic), but first one was already there

I can easy reproduce this issue, but still do not know how solve it and what actually cause this. 

eg if it may be solved with padding setting etc.. or I should not use split vtx normals... or if there is another user preference setting which related this matter.

I tried change resolution, or change UV island expand for some non painted parts.  (divide UV set to separate island (UDIM) may solve some issue. )

and remove seam too. (but keep same UV island and return to paint room)  etc.  . so I think some part can be painted now. 

at same time, when you hide all layer of this robots mesh,, it sitll remain some float verts. it never disappear from paint room untill I remove the object.

  so I think the geometry related too.. ( though I know this robot is offered to learn paint, and most of case it is good enough if I use in render scene.)

but if it had worked, then change with versions, I hope to know the reason.. so I can avoid it when I see same issue for painted mesh...  

Share this post


Link to post
Share on other sites

Merging all under the same island (or something with minimal number of seams) is not accurate when you deal with normalmaps/baking, seams allows to reduce polycount and avoid too-thin triangles which are a pain to render for your graphic card and also may cause extra-checkup work when fine-tuning skin weights  (even with a strong tool like Akeytsu)
To me actually the solution is simply to downgrade at a lower version, as said above (4.8.44, I wish the server could display the full history but I could only retrieve this build thanks to the link posted in 4.8 Beta testing thread), I hope the bug will be solved one day but this old version still does the job so it's fine to me actually (I don't want you to stop your plans with the next huge 3DCoat-2021 which I can't wait to play with )

Cheers Pilgway team, you're the best !

Share this post


Link to post
Share on other sites

I understand basically I need to set seam untill unwrap then locate each island in UV map.

Then what I did not know,, I sometimes see when I import mesh with UV,, there is case,, it is already un-wrapped but I can not find any seam as visuall.

And there is case, to use select island, I need to set seam again in 3d coat.  Then about this case,, It seems reduce problem if I remove seam only.

(that means remain the UV map as same as before )  but I do not know, to keep UV but remove only seam is good aploach for baking etc.

Will it offer difference for quarity ?  or 3d coat may need to keep all separated edges (which used when unwrap UV) as marked "seam", to bake texture correctly?

(it means,, we need to keep seam even though you already get unwrapped UV,, or we need not to keep seam if I do not edit the UV any more)

Edited by tokikake

Share this post


Link to post
Share on other sites

That seam issue is a bug all ready fixed on new version to release soon.

  • Like 1
  • Thanks 1

Share this post


Link to post
Share on other sites

18.11.20 4.9.66 (stable)

- fixed FBX UV-set importing problem (bot usual, rare problem).

- HSV control corrected to achieve true black color if need.

- If the pen created from the TIFF/PNG/PSD the zero-depth will be taken from the corner of the picture instead of using RGB(127,127,127)

-  Fixed symmetry and stencil in paint tools.

- Symmetrical copy corrected.

- Opacity in color map taken into account in smart materials.

 

Seams problem not corrected there, it is a bit more complex story...

I am mostly devoted to 2021 development now.

  • Like 6

Share this post


Link to post
Share on other sites
On 11/18/2020 at 7:50 PM, Andrew Shpagin said:

18.11.20 4.9.66 (stable)

- fixed FBX UV-set importing problem (bot usual, rare problem).

- HSV control corrected to achieve true black color if need.

- If the pen created from the TIFF/PNG/PSD the zero-depth will be taken from the corner of the picture instead of using RGB(127,127,127)

-  Fixed symmetry and stencil in paint tools.

- Symmetrical copy corrected.

- Opacity in color map taken into account in smart materials.

 

Seams problem not corrected there, it is a bit more complex story...

I am mostly devoted to 2021 development now.

Thanks I could confrim Opacity in color map taken into account in smart material.

at same time at current once I set alpha include texture for smart material, now smart material perfectly overwrite which already paint on same layer.

then the opacity area remove the color which alredy painted in the layer. (I do not clear remember it is expected or not,,)

Though I can manage it with use different layer... of If there is other setting, which not remove painted area but add the image on to the paint ?

1316187479_removecolor.thumb.JPG.d3740c4704a3fe13db0861c789d764b6.JPG

as you can see, I already paint base on the layer 2, then try to put decal (smart material which set png as color) on to the base.

now it remove the base paint area and make it opacity..  (to clear see I hide layer 0) .:mellow:

And thanks, now I can paint without tiling  ^^

Edited by tokikake
  • Thanks 1

Share this post


Link to post
Share on other sites
On 11/18/2020 at 11:50 AM, Andrew Shpagin said:

Seams problem not corrected there, it is a bit more complex story...

I am mostly devoted to 2021 development now.

It's fine, I'm okay for texturing using 4.8.44 and I will keep an eye on new builds, appart from this issue 3DCoat remains a solid tool and I'm super excited to see what will come up in the 2021 =D 

  • Like 1
  • Thanks 1

Share this post


Link to post
Share on other sites
On 10/6/2008 at 11:47 PM, Andrew Shpagin said:

- HSV control corrected to achieve true black color if need.

I feel like for once I've influenced the course of 3D software, even if it was by my ineptitude at making Emissive maps! hahahaha

  • Haha 2

Share this post


Link to post
Share on other sites
On 11/13/2020 at 11:23 AM, Vexod14 said:

Hi everyone !

It's been a while since I didn't report anything here, I'm having an issue in painting room around seams, as depicted in this video :




Here are two big problems :

1 - painting over seams is no longer reactive, since you often have to hide then show back again your active layer to diplay what has really been just painted
2 - some pixels remains definitely "unpaintable", which causes serious problems when you need to texture something...


I'm usually running the GL version, but DX fails the exact same way.





 

I think this is the same kind of issue I've been having for the last couple years that i try to ask for help about.. but for me in those areas I'd likely get spikes and i think it's inverting the padding/black area on the UV..  In my observation the brush will get confused from jagged UV seams or topology and cause the seam/padding to expand .. in your case the missing color might be that you're trying to paint on the expanded padding.  I may be wrong about what is causing your issue but it sounds the same. 

@Andrew Shpagin I've updated the conversation about this with you and Carlosan in my inbox. The example as you say may be extreme but these spikes happen randomly, maybe one or many more show, often after after hours or days of work.. you don't see them right away unless you use materials with a  lot of depth. So I hope there is a fix for this soon.

Edited by Ascensi

Share this post


Link to post
Share on other sites

Sorry it is not critical, but when I save "scenname.3b" file, (3d coat scene) in another hard drive directory, (in 3d coat user data directory)

3d coat can save the file (of course) but recently I often see message "unable to write file ".../.../ scenename.3b.jpg"

like this. It seems when I overwrite same file, 

notgenerateimg.thumb.JPG.12de844299390c989430ef31be2a4629.JPG

 3d coat not generate current scene image. (so not up-date scene image file, but can overwrite current scene 3b)

if it caused by folda permission issue, I think it may deny to save 3b file too. so it seems there is tiny problem , when save (overwrite) 3b scene with  current scene image.

 

 

Share this post


Link to post
Share on other sites
55 minutes ago, tokikake said:

if it caused by folda permission issue, I think it may deny to save 3b file too. so it seems there is tiny problem , when save (overwrite) 3b scene with  current scene image.

 

 

3d Coat can indeed overwrite the preview image. Do this :
-save a file
-Save incrementally (ctrl+shift+s)
-close file
-reopen original (the file saved in step 1)
-Save incrementally  >> both the .3b file and the .jpg are overwritten without a warning message, nor an error.
 

Share this post


Link to post
Share on other sites
45 minutes ago, Silas Merlin said:

3d Coat can indeed overwrite the preview image. Do this :
-save a file
-Save incrementally (ctrl+shift+s)
-close file
-reopen original (the file saved in step 1)
-Save incrementally  >> both the .3b file and the .jpg are overwritten without a warning message, nor an error.
 

Thanks I will challenge it (though I do not think I hope do same steps to overwrite scene image) and I feel, though I see error message, but actually if 3d coat overwrite image , I did not test it.. anyway if it happen for all user, I may not care ^^; (but report for enhance)

but when I check folda, time stamp seems old so I suspect if it not overwriet scene image.. (basically i simply believe " I can not do it"  when info window tell me, "No you can not"  

Share this post


Link to post
Share on other sites

24.11.20 4.9.67 (stable)

- fixed particular seams problem - painting over geosphere produces seams in the middle of the big UV-island.

Share this post


Link to post
Share on other sites

×