Jump to content
3DCoat Forums

farsthary

3DCoat Developer
  • Posts

    276
  • Joined

  • Last visited

Everything posted by farsthary

  1. Hopefully this beta will be released soon, Andrew is interested as-well to release it. Currently I have several known memory undo bugs, so if can't solve them before beta release, use it with caution and saving the model often.
  2. Thanks. They eye socket test was a quick sketch to show the looping. Can somebody sent me a head model to test on my own? My modeling skills have grown rot over time :p
  3. @ajz3d inner strokes are not frozen , they can be smoothed a little, but it is not difficult to give the option to fix them.
  4. Also for spline like patches will come QuadStrips. The outer contour will be defined with easier to control splines and the inside will be filled with this providing even better quality.
  5. Currently it doesn't work with selection shapes. You have to stroke them. Is interesting what you propose but other than the initial patch I don't see many uses for it. As you start filling the mesh you dont have many choices to create perfect ellipses, rectangles, etc. You just draw what you need according to the remaining surface and the boundary of existing mesh.
  6. @gball with hybrid you can create loop like but not using concentric loops, using 1 loop cut in half by one stroke, quads will follow the outer loop. A quick test using only DIRECT, it require more tweaking than hybrid though.
  7. Yes, the DIRECT method is the "loose" retopology you are talking about, is nice and clean, ZERO poles, at the expense of giving up some control, like polyxo and AbnRanger said. I should probably demonstrate it more. The purpose of the HYBRID method is that the patch shape dictates the polygon flow and there lies the control, you don't need HOURS to came up here with nice results, just minutes, yes, sometimes you need to tweak afterwards, but those are minor tweaks. You can experiment , go back and forth and for artist you will master the tool very quickly. This is a semi-automatic retopology tool, it means you will still have to work for what you want, is not a button do-it-all-for-me. Eventually I could work in a fully automatic method, either improving Autopo or extending it with these algorithm but for now I'm here. In those clips I showed careless patching, for demostration purposses. A real artist who wants to control the polygon flow better will come up with better patches. Direct: patches loosely controls quad flow, No poles Hybrid: patches strongly controls the flow, some poles but more natural flow according to the parch shape.
  8. @Taros QuadStrips works very similar to Blender's RetopoFlow. Is spline based quadrangle strips, instead of using just strokes like QuadPaint. It will be easier to outline patches and then filling them. I had to pause it in order to finish QuadPaint because the last one is the main algorithm and was the most advanced, but should quickly follow QuadPaint first public release. It has no functionality yet in the public release, is an alpha there.
  9. @Tony Nemo RFill got several bug fixed recently but I haven't merge yet, RFill, QuadPaint and QuadStrips share big portions of the core so improvement in any of them reflects in the others. @Manticor Yes, That's a last stage for another tool (or an option of QuadPaint) in fact it has been in my background long time ago this possibility.
  10. More fixes and updates. I've made that really quickly with few strokes.
  11. @TimmyZDesign It is very interesting your idea, thanks! Currently QuadPaint algorithm does not work on poles as an input, they are just the output. I will currently take the algorithm to its full potential as is also a research in progress too. For future iterations it will evolve much more but for now I need to focus on ironing it, so the huge base of artists can start using it and feedback me with field data. I haven't merge my code with Andrew trunk so probably the version available in public builds is way behind, more in the alpha stage. Will do it soon! Cheers
  12. @Carlosan Yes, double click and you have a new pinned vertice
  13. @AbnRanger and Michaelgdrs True, if the contour has an odd number of vertices it will have at most 1 triangle (quad dominant), but if is even it will be quad complete. In the video I was showcasing a somewhat sketchy retopology where I didn't care about the evenness of the patches. The good news is that you don't have to select strokes in any order and you will be able to control that with vertices insertion/deletion, splits, cuts, etc. Is already implemented into that SimpleStrokes sub-tool but not just exposed yet. Also a preview mode will help to work in a non destructive way before applying to mesh and showing each patch side vertices count will greatly aid in that. With even strokes and matching vertices the quality will be very, very good and moreover, Is fun! I actually find myself playing with the tool drawing crazy patches and explore different mesh partitions easily. This method can be extended also to Autoretopology if I implement auto partition of the mesh with patch pairing and it will have the advantage of not shrinking too much polygons in small areas like the current Autopo, but that will be another tool Imagine also this tool used in combination with QuadStrips where you have more control on the splines. with QuadPaint I want artist freedom.
  14. Hi @Haikalle Thanks, I will check it today. Yesterday I also improved the new quad filling algorithm a lot
  15. I've solved the mesh explosion bug in Undo. Andrew will commit it soon
  16. @haikalle hmm.. couldn't find the issue even on your testbug file. http://recordit.co/y2CilX9cmB As an extra note there are some tools, like rapid and rapid2 where is possible for mesh to become stretched and overlapping and if you continue to sculpt over stretched triangles or overlapping mesh of course it will grow into ill mesh, those tools by definition don't protect the mesh from the artist, in those cases small sanitation as you sculpt is advised, like smoothing here and there, may be an undo... that's not the type of bug I'm looking, but rather mesh "explosions" caused by inverted/uninitialized normals as show by digman picture.
×
×
  • Create New...