Re-topology is there anything on the horizon?
polycounter lvl 12
So much has changed but seemingly re-topology has stayed relatively the same for the past several years, perhaps it is just more noticeable for me as its a part of the process I find tedious, some people I notice have had similar gripes with creating uvs.
for ideal optimized geo retopoligizing by hand seems the only way to go (doubly so for non organic surfaces) as auto retopology tends to fall painfully short in my experience and is usually as far as I can tell just being used as a cleaner base to retopo from, doubly so when it comes to hard surface.
I was just curious towards the rest of your thoughts on it do you think it is "stuck in the stone age" compared to most other advancements in workflows and processes, do you see some kind of a major advancement on the horizon? or have you seen one already lol in which case please share!
for ideal optimized geo retopoligizing by hand seems the only way to go (doubly so for non organic surfaces) as auto retopology tends to fall painfully short in my experience and is usually as far as I can tell just being used as a cleaner base to retopo from, doubly so when it comes to hard surface.
I was just curious towards the rest of your thoughts on it do you think it is "stuck in the stone age" compared to most other advancements in workflows and processes, do you see some kind of a major advancement on the horizon? or have you seen one already lol in which case please share!
Replies
Still waiting for an update but I think retopo will be in the futur 100% automatic ( I hope so :d )
Just check this video out around 3:30..
And there are hints that there might be some Retopoflow demos at Siggraph this year.
Regarding Blenders UV tools;
watch this video called; UV unwrapping gets an overhaul
there is also a UV packing addon for Blender available
https://gumroad.com/l/UVShotPacker#
and some nifty custom optimizations for the UV editor;
https://gumroad.com/l/Smart_uv#
Grimwolf
please be more specific on what you would like to get improved in Blender.
The Blender Community is open and listening and improving Blender on a daily basis.
my current workflow though is generally zremesh with guides for the broad areas, then take that into 3d coat and delete all the joint areas / face and do them manually for anything deformable that isn't a hero asset (which would get done fully manually). Static objects just get decimated unless there's a good reason not to.
Currently what I have is a very solid base mesh that I can use at subd 1 or 2 depending on polycount and call it a day. I haven't had to build hand or face topology from scratch in a while. Just finish sculpting and export out the parts I need.
We're hoping to demo an early alpha version at SIGGRAPH next month. The main thing we're doing is unifying all tools into a cohesive retopology mode, making retopology work more like sculpting than edge modeling.
We've never stopped working on it; progress is just slow as it's a side project.
cool to know those details
Some features where able to keep already unwrapped areas, is it still possible to keep already unwrapped areas when using the unified retopo functions?
Another question;
Any chances to see Edge patches gets added or is included in it?
the auto-retopo solutions i've tried don't satisfy my clean-mesh-OCD and quad-draw over surfaces seems inferior to projecting proven topology so i'm not even looking at fixing retopo.
the only solution i use very occasionally is 'instant meshes' for quickies.
http://www.russian3dscanner.com/
Obviously it makes sense for studios to use Wrap but lone artists who already have ZB and are only doing this work every so often?
And you won't just be limited to 'wrapping' polygons
I regularly use Wrap and WrapX and can only recommend them.
Combining Wrap/WrapX with Retopoflow and other plugins/programs is a sure way of finishing the Retopo step much faster.
Here is an extreme case I did recently with Wrap. (please excuse all the spinning)
Some manual clean-up for relaxing stressed areas is still needed but any program which has a brush and re-projection function available should be able to do it. For that task I tend to use Topogun or one of the first versions of Wrap which had a Brush function.
First I tried wrapping a headscan of my own(1mill tris) to their default head basemesh. The sim ran so long that I had to leave it overnight and this morn it had crashed.
So I went to a simpler example and used their neutral hand scan and base mesh. The sim ran about 3 hours and had only gotten half way through the 3rd (final) subdiv pass. At this point my enthusiasm had dwindled and I shut it down (through task manager as apparently it is an uninterruptable process)
Wrapping node settings at default. I can't find any info on min sys requirements on the site or google. My workstation is not available so I'm using a latest gen i7 laptop with 16 gigs ram on win10. I thought this machine could surely handle the most basic sim using their own assets?
Can you relay your own experiences on this? Cheers.
It is recommended for quick computations, that your basemesh has not more than ~30k polygons since the more polygons your Basemesh has, the longer the computation time is.
According to the developer, a 100k basemesh might take up to half an hour of computation time.
Your scan/sculpted mesh should have around max 3-5 million polygons.
Wrap/WrapX makes heavy usage of the CPU,the GPU is used when you have Nodes for Texture transfer.
The Rigid Alignment Node is needed when your Scan /Sculpted Mesh is to small or in a strange Angle and allows the shortening of computation time once your Meshes are aligned.
Using the Select Points Node wisely can massively shorten the computational time.
It usually takes only a couple minutes of computing time on a desktop i7-3770.
Existing UVs will be maintained, no worries there. In 2.0 we're writing directly to Blender's edit mesh, using the BMesh API. The one thing that likely won't be supported is translating Vertex location changes over to the UVs while modifying via RetopoFlow. It's likely possible but way out of our scope.
Edge Patches in it's current form likely won't make it in but we're aiming to take the basic idea and integrate it within the 2.0 architecture. So you should see something like it, to a degree!
If you're interested in 2.0 then check out my twitter feed, I'm posting GIFs as we make progress. I may make a dedicated thread for it once we're a bit further along.
Maya also has quad draw which you can use on a live surface (your decimation zbrush sculpt).
You can also topo using Zremsher lines (it's a brush) to control zremesher better. And for working on hard surfaces, the workflow preference is breaking up sections where possible (seam lines, extrem angle changes) and zremeshing that part of the mesh.