Page MenuHome

Retopology in Blender
Open, NormalPublic

Tokens
"Love" token, awarded by Frozen_Death_Knight."Love" token, awarded by A.Lex_3D."Love" token, awarded by michaelknubben."Love" token, awarded by FaisselB."Love" token, awarded by johnsyed."Like" token, awarded by belich."Love" token, awarded by lcs_cavalheiro."Love" token, awarded by Tonatiuh."Love" token, awarded by koloved."Love" token, awarded by TheFlow."Like" token, awarded by Jaydead."Love" token, awarded by ixd."Love" token, awarded by erickblender."Love" token, awarded by EitanSomething."Like" token, awarded by mendio."Love" token, awarded by 0o00o0oo."Love" token, awarded by Jules."Like" token, awarded by amonpaike."Love" token, awarded by Zino."Love" token, awarded by Alrob."Love" token, awarded by xrg.
Assigned To
None
Authored By

Description

In Blender’s overall workflow, which includes modeling, sculpting and animation, a major missing piece is retopology.


No Retopology?

As part of the overall focus of sculpting improvements, this is an area we should address.

After talking to @Pablo Dobarro (pablodp606), we agreed to set up this document to help define how we should approach this from a UI and workflow perspective.

Issues

Display

One of the major issues currently, is that we have no great built-in way to display your retopology mesh on top of a high res sculpt.

By default, your mesh will intersect with the high res sculpt:

You can enable In Front, but that makes the *whole* retopo mesh display on top of the reference, also the parts that should be occluded:

Not only is it confusing to set it up, but the result is sub-par.

What we really want is something more like this:


The occlusion is correct and the faces aren't intersecting. The retopology mesh is half transparent so you can see the underlying detail.

Snapping

There are a few issues with snapping:

  • You have to enable some very specific settings that are not obvious
  • There's no way to only snap to a certain object or Collection - you can easily accidentally snap to an item somewhere in the background
  • Snapping is only active while you are using any of the transform tools. If you for eg subdivide an edge, those edges aren't snapped to the source. Users then have to select those items and move them a bit to snap
  • Some users use the Shrink-wrap modifier, but this has issues with the mesh is too far away from the source - the projection becomes unpredictable
Tools

Compared to more dedicated solutions for retopology, Blender is lacking many basics. Things like drawing a 'strip' of quads, or defining a patch of quads. Or tools to automatically handle appendages just by drawing lines along it. Or even just a basic tool for 'drawing' topology onto the source mesh, which is then auto-filled with quads.

Setup

To get started with retopology in Blender, so many settings and display options have to be toggled and set by the user, that there exists entire tutorials on how to do this. The needed steps and settings are not only non-obvious, but also take time and focus away from the task at hand.


Solutions

Display

Add a Retopology overlay display option, replacing the current Hidden Wire Edit mode display

Snapping

Add a Retopology snapping option, which allows snapping to self while also projecting onto source, and snaps the current selection.

Tools

We can start by improving certain tools, such as Poly Build, and also add more edit mode tools. Even though they may be more constrained inside Edit Mode, we can try to use this as a test-case for more advanced gizmos and tools in general

Other examples of the kinds of tools we could add are things like:

A tool that lets users define a patch with handles, which is automatically filled with quads:

A tool that would let users draw a line, which then becomes a strip of evenly sized quads

A tool that would allow users to nudge around points with falloff, like a brush:

Addons

We probably don’t expect to provide a full and comprehensive set of retopology tools, at least initially. For this reason, addon developers could provide extra tools that integrate with the toolbar and tool system.

Details

Type
Design

Event Timeline

There are a very large number of changes, so older changes are hidden. Show Older Changes

I also think that it is better to have retopology in a separate mode. Adding this kind of tools to edit mode will require them to control the selection mode, overlays and scene snapping configuration. The UI and the code are going to be a mess.

You should be able to switch from retopology to edit mode instantaneously. It is not like switching to sculpt mode or from object to edit mode where you need to build new data structures to start editing the mesh.

I'm not going to put myself in either camp, but if you're going to make a new mode, please give some thought to this important consideration:

While it is true that generally in retopo you want to create a mesh that adheres to the surface of another, this is only true in a very broad, general sense. Besides sticking to the surface, retopo meshes must also not be messy so that they are easy to unwrap, weight, and animate. Particularly for environmental assets, the meshes often must be neat and strictly conform to grids or measurements so they can easily be snapped together, and this is much more important than sticking to the highpoly model. This is where standard edit tools, with their freeform placement of vertices come in.

It's clear that if the new retopo tools are a new mode, the workflow will involve very frequent switching between modes. To facilitate this, such switching must be quick and painless. The display options must be able to be made identical between edit mode and retopo mode so that there's no extra fiddly setting switching or disorientation whenever you change modes - in essence the retopo display options must be independent of the retopo toolset. There must be no warnings or lost data layers while switching.

And most importantly, retopo mode must never, under any circumstance, automatically snap vertices that were deliberately placed not-on-surface by the user.

While reading comments i had the feeling that we are talking about different approach to retopo.
There are different ways to retopo: Autoretopo, Manual retopo and Semi-manual retopo

Auto-retopo: we have in the sculpt branch "OpenVDB remesher". Some sculptors can be happy with this approach. After sculpting they can throw the remesher and they are happy. They don't care much about the face loop flow. And some users want a better quads base autoretopo.

Manual retopo is when the user has full control over where he/she wants to put the vertices, this workflow is needed when you don't have good base mesh and after retopo you will use your model for animation. To make this process much easier Blender has some free addons ( bsurface, mira, retopoMT, F2, ...) Here is a video of me doing manual retopo while Blender was in earlier beta version. https://youtu.be/OdeuxXe9V6A

The Semi-manual retopo is what julien had mentioned above: the user has a kind of base mesh, uses shrinkwrap modifier and from time to time goes to edit mode and sculpt mode to tweak the vertices positions.

Questions to ask is what are we planning to implement here, all the 3 ways to retopo? Code wise what is the best approach? Can we activate a mode like sub-mode of edit mode like carverMT, retopoMT are doing right now? Looking at my old video, i can say that there is so many options to activate before starting the real retopo process, automating all this will speed up the workflow. For me i m open to all the options and i like to test things in practice to understand them better.

Would this retopo mode have selection and selection tools at all? Or would it be more a paint type mode, where you draw / erase / nudge edges and faces without selection?

This comment was removed by Zino Guerr (Zino).

The mesh display in the third picture is not very good either as there is no way to judge at a glance whether any given pixel is above or below the surface of the source mesh. I'd suggest at least drawing an outline of where they intersect, or maybe drawing the portions above the source mesh in a slightly different shade.

Also regarding Workflow point 4, source "objects" only rarely consist of only one actual object. So storing the retopo mesh reference per object is not great. Instead maybe store source object collection per retopo mesh?

@Brecht Van Lommel (brecht) I think each tool should manage its selection state according to its functionality instead of using the current selection tools. For example, a grid fill tool can store the edges needed for the operation without modifying the current edit mode selection, and then notify the mesh drawing code with elements need to be highlighted. This kind of tools should be the exception, most of the tools should be designed to be more like brushes, without needing a selection at all.

@Piotr Adamowicz (madminstrel) Yep, you should be able to define a Collection, not just an object, as a reference target. As you say, sometimes you may need to retopo something that consists of several objects.

What annoys me with this idea of new mode is that described tools would be as useful in other fields without use of an automatic snapping.
I can imagine using Patch tool to create clothes or terrains.
Drawing directly a strip of evenly sized quads for same purpose as Draw Curve is currently used instead of drawing a curve, converting it to mesh, extruding it.
Tool that would allow users to nudge around points with falloff, like a brush for shapekey editing of a character using a muscle system.

Fundamentaly, Blender would become more powerful if those new tools were not restricted to one unique use.

Adding this kind of tools to edit mode will require them to control the selection mode, overlays and scene snapping configuration. The UI and the code are going to be a mess.

If the tool implies a preselection highlighting, selection mode should not be more problematic than current Loopcut tool.
I don't think that tool should control overlays and scene snapping. That should be user choice.
@Piotr Adamowicz (madminstrel) also talked about the need to let snapping control to user.
We would like more overlays for Edit mode to clean up meshes without automatic snapping.

And having a workspace dedicated to Retopology with appropriate Overlays and snapping configuration should be possible without creating a new mode.

I understand that a more powerful Edit mode is adding more work and complexity to developers than just keeping things at level of a more limited retopo mode.

To me, this idea of a retopo mode looks like a missed occasion to improve Edit mode, a missed occasion to improve customization of Workspace and an enterprise with a limited goal that would delay such kind of achievement.
New retopology mode or not. Users will continue to ask for a better and more powerful Edit mode and UI.

I agree that it should be part of Edit mode. At certain point meeting halfway between what it's easier for the developer and what's best for the user steers too much into the developer side. The same way flat design is popular because it's easier to produce and mantain, not because of the end user.

What could be done is:

-Add an option to choose a collection to constantly snap to from all directions like shrinkwrap, not just the view. Some tools could access this as well
-Add the drawing mode above, it could work like this addon, or it could access the target collection to determine depth somehow


-Implement those same tools from above, which are useful for non retopology.
-Maybe make the F2 addon a default
-And add a workspace for all of it

I do a lot of retopology in Blender with just F2, shrinkwrap, and the default tools. Not at any point have I thought "Hmm, I wish to not have access to these tools while I retopo" Retopology sounds too specific of a word to dedicate an entire mode to it, particularly when handling the same data. It's like bending backwards to the workflows the industry says there only exist, instead of creating your own workflows, by mixing and matching tools. The old UV Editing mode from 2.4 honestly made more sense than this

Other modes in Blender do suffer from this, but at least they are different data types so they have an excuse. Sometimes I wish to extrude a new key with E in Particle edit mode, or subdivide hair keys, or use many of the selection commands from edit mode in Weight painting mode. It is a problem and making new modes only makes it worse

I mentioned this when chatting with @William Reynish (billreynish) and I think it's worth noting here as well:


In sculpt mode you have dyntopo. The idea is that certain tools like "simplify" will only be available in the toolbar when this is activated, since they wouldn't work otherwise. I had a similar idea with activating painting in sculpt mode and then you get color options in the tools.

What if there's an option in edit mode to enable "advanced snapping" (just coming up with temp names).
You enable that option and you get all the important options to target other objects/collections for improved live snapping.
When this is enabled, the tools, like drawing geometry around limbs, would become available, while other new tools like patches, strips and push/relax brushes would automatically snap like the other tools.
Then you have everything in one place without duplicating tools between 2 modes or no need to switch between an edit mode and retopo mode to access different editing tools.
The rest is overlays options for better display of the mesh on top of other objects.
A separate mode or the "advanced snapping" option are ambitious changes either way.


I do get that it would kinda create a "mode within a mode" but it would also make all overlays, tools and workflows available in edit mode at the same time without splitting them completely and possibly making different workflows harder.
I think the benefits of creating a retopo mode are just as big as the potential problems of gating of tools from edit mode and vise versa.

If we want to change the way we tweak the geometry we can improve the already existing Tweak Tool. Same for polybuild, hidden wire overlay, snapping and any new tools.
Now that overlays, tools & workspaces are a part of Blender it would be way easier on a user level to have it all in edit mode.

I think @Erick Tukuniata (erickblender) brought up a fair point though. The cost/benefit of a separate mode strongly depends on the level of automation we can expect from the new tools. If the target is excellent manual retopo, then going with improvements to Edit mode would be much preferable. If the devs are more ambitious and are going for something semi-automatic, like described in Data Driven Interactive Quadrangulation, then a new mode would definitely be the way to go.

I imagine a new workspace could be enough. But of course, always in on other ideas to rethink. :)
"Retopo" tends to make me only think about re-doing already existing models. While I think a lot of usage could be simple drawing out of a scene or building precise models besides retopo. If a workspace will be it, it should probably just be called "Topology".

Right now, one thing I think is most important:
For this we need the simplest tools, like pen on paper. Maximum control.

  • Building vertex by vertex, exactly where the user wants them with automatic filling (tris, quads, ngons). Also with the options to snap on surface, snap to grid and a symmetry mode.

How about the PolyBuild tool? Could be a great start for the development of those new topo tools since I think it already does half the needed functions.

The "move" brush is also a good idea, could maybe be combined with relax by holding shift.

PS: And whether the tool can create new objects, or "Add Vertex" from the "Extra Objects" Addon should be included out of the box.

I think @Erick Tukuniata (erickblender) brought up a fair point though. The cost/benefit of a separate mode strongly depends on the level of automation we can expect from the new tools. If the target is excellent manual retopo, then going with improvements to Edit mode would be much preferable. If the devs are more ambitious and are going for something semi-automatic, like described in Data Driven Interactive Quadrangulation, then a new mode would definitely be the way to go.

I think that's a good point but I can imagine something like this to even be its own object type with own dedicated modes. This is also much more ambitious and definitley outside of a manual retopology workflow.

Re. the mode discussion:

Part of the issue, is that some tools may *require* a snap target to work, meaning that some tools may need to be disabled under certain circumstances, if you don't have snapping enabled. I'm not sure some of the kinds of tools we have in mind would be able to work in regular edit mode, but perhaps @Pablo Dobarro (pablodp606) has some additional perspective on this.

Same for the fact that selection and snapping may be handled differently, although it may be possible to make changes to snapping to support only snapping to certain Collections.

Also, more on a practical level, the fact that currently, in Edit Mode, it requires a lot of setup by the user to get started. An > 6 minute tutorial exists on setting it up.


That said. we could probably also also do a less ambitious solution, without so many specialized tools, and just improve the display, and do some tweaks to snapping and then leave it at that. Separately there's a possibility that we could begin to add some more tools to Edit Mode, like a Quad Paint tool for painting strips of quads, a Tweak Brush tool for nudging vertices around, and so on.

Then, assuming we could do that, the issue remains that setup is still very complicated, which would then have to be solved in some other way. In theory there could be an operator you could run which would go and set the necessary options, although 1, you'll need to know to use that, and 2, then your settings will be all wrong when you are finished the retopology process.

The toolbar UI ideally also would need to perhaps become customizable, or there could be some way to create sets of tools so that retopology tools wouldn't drown below the regular Edit tools.

@Piotr Adamowicz (madminstrel) Yes, that is partly it. I imagined that many of the tools would operate on a higher level than on verts, edges and faces. Some retopo solutions do this, so that you focus more on the topology, rather than the density of individual points.

Although there's also a chance that an active tool could do this kind of thing.

That's a good case for a separate mode then. Though ideally you would then want to have that higher level remain persistent rather than disappearing as soon as you finish the operator or switch out of retopo mode. For that it'd probably be nicer to have a whole new object type that could store all the bezier handles, segment counts, densities, manually placed poles and so on.

@William Reynish (billreynish)

That said. we could probably also also do a less ambitious solution, without so many specialized tools, and just improve the display, and do some tweaks to snapping and then leave it at that. Separately there's a possibility that we could begin to add some more tools to Edit Mode, like a Quad Paint tool for painting strips of quads, a Tweak Brush tool for nudging vertices around, and so on.

Then, assuming we could do that, the issue remains that setup is still very complicated, which would then have to be solved in some other way. In theory there could be an operator you could run which would go and set the necessary options, although 1, you'll need to know to use that, and 2, then your settings will be all wrong when you are finished the retopology process.

I don't think the setup needs to be complicated. So what would the setup within edit mode probably be to have a closer look:

  • Enabling the targeted snapping and selecting a target object/collection.

This could be as simple as clicking on a toggle and picking an object/collection, probably similar as manually doing in in a retopo mode.

  • Enabling overlays to make the retopo-mesh better visible

These could be a couple toggles in the overlays but as with other tasks with lots of setup: The Workspaces can save a lot of time there. For sculpting I would also first go into overlays & shading option and disable/enable a bunch of stuff and workspaces save me that time.

Any other setup like adding a build in mirror option or the mirror modifer would be almost the same time investment. Same for a subdiv modifier.
What else is there? IMO this doesn't seem to be complicated. In the current state it's a notable time investment because of the visibility and snapping issues and in the 6 min tutorial, over 2 min are dedicated to explaining the setup, but with some improvements to the overlays and snapping I don't think the "complicated setup" is what justifies the new mode.
The real issue from what I heard so far would be making the new tools work just as good in edit mode as when they would get their own mode.

@Pablo Dobarro (pablodp606) @William Reynish (billreynish) named some important issues and I would love to know more:

Adding this kind of tools to edit mode will require them to control the selection mode, overlays and scene snapping configuration. The UI and the code are going to be a mess.

The selection mode switching based on the tool is a good point and it's already happening in edit mode with the polybuild tool. Try to use it and it will enable vertex selection mode. If that's a good thing or a bad thing I'm still on the fence about.
I don't know which tools would need to override the overlays configuration but when they need to display some extra overlays they could likely just be part of the tool gizmo. The snapping configuration is important for a couple tools so that goes back to the "mode within a mode" solution/issue.

If we want complex semi automatic retopology tools, brushes and so on, we will want to have a new internal API that makes the development of these tools as easy as possible. Let's say we are designing a patch tool. If we want to add it directly to edit mode, we need to code it in a way that supports all element selection modes, every combination of overlays and shading modes possible, it should share the same keymap as the rest of the tools from edit mode, it should not modify the selection of other elements, it should support all combinations of scene snapping modes and targets... when in the end, 99% of the time, you will want to use this tool for retopology, with the same overlay, snapping mode and without worrying about selections and drawing modes. It is going to be an insane amount of work and bug fixing just for these corner cases.
Some of these tools are hard to code even in a clean environment that should provide a separate retopology mode. Just getting the snapping, symmetry and the viewport visualization right is a huge challenge.
Maybe some of the tools don't cause any problems and can be added to edit mode as well, but more complicated tools like quad strips or contours need an extra layer of abstraction that should not be designed with all the features from edit mode in mind.

@Pablo Dobarro (pablodp606)
I get that it would probably be easier to develop but it doesn't sound like the rules are that strict like you say.

we need to code it in a way that supports all element selection modes, every combination of overlays and shading modes possible, it should share the same keymap as the rest of the tools from edit mode, it should not modify the selection of other elements, it should support all combinations of scene snapping modes and targets...

The shading options are still going to be a thing in every mode anyway and most of the overlays are shared between all modes.

The selection modes and "not modifying selections of other elements" doesn't seem to be enforced anyway, since using different edit mode tools already switched selection types and discards/updates selections.

For the scene snapping modes it isn't even that strict either. Try to use some tools with snapping and they just won't do it. Objects/Collections as snapping targets would be an option that would be taken into account anyway and that other edit mode tools could benefit from too.

The keymap is something I totally get but since a lot of shortcuts were removed for 2.80, there should be room. Otherwise there's always a spot in the toolbar and for custom shortcuts and we still need a dedicated "tools keymap"

On another note: If we also add half of the retopo mode tools 1:1 to edit mode and maybe even more options, then the split will seem very arbitrary. Going the other way and making tools exclusives to specific modes can be very frustrating though.

Then maybe we don't want those tools at all, or speaking for myself, I don't. They sound so specific that they're more fit to be part of an addon than anything else.

The cornerstones of retopo IMO are the Overlay and the Snapping, just those two things. Not tools that may or may not take advantage of said snapping. It is what I wish for every time I'm doing retopo. But I'm never wishing for patch tools, I can handle polygons, vertices, patches, faceloops, all that and their manipulation just fine.

After trying many retopo solutions, patch based, curve based tools are never faster for me than my experience manipulating vertices directly, the only really useful one is the contour tool, and I rather that be an addon than break Blender into modes and/or bugs. Then there are fully automated ones like zRemesher, that's actually useful. It's that middle ground of semi automatic tools that, while useful, I don't find useful enough to make such drastic decisions over.

@Wo!262 (wo262) Even though it kinda sucks to say it: I think you have a point.
Calling these new retopo tools "basic functionality" is not accurate IMO. The most important additions are better overlays to see geometry while retopologizing and better snapping options. And ideally all of that within the new slick interface of 2.8 to get rid of time wasted for setups.
Most new tools would be part of edit mode anyway like the push/relax, updated grid fill & polybuild, and maybe even more. Tools like contours, polystrips and strokes will be super helpful but are these couple tools worth it to split Blender into more modes?

And it's disheartening to say that we probably should go for the "minor improvements to edit mode" instead, but these are the big improvements! The ones that really matter.
And there's likely still a way of implementing the more complicated tools in edit mode, even if it takes more time.

Agree with @Wo!262 (wo262), currently, the main issues to do retopology in Blender are snapping and mesh display, most of the tools we have in edit mode are great for the task, we just need some extra polishing on some of them (like having new vertices automatically snap to the source mesh after a grid fill for example), but for manual retopology a new mode sounds like an overkill. The semi-automatic tools mentioned by William and Pablo sound good to have but not terribly necessary, and for fully automatic retopo like ZRemesher we could use addons and or an improved version of the remesh modifier.

If the snapping is improved that could benefit other modes too, same with more overlays options, so I still don't see an entirely new mode as something really necessary.

for those who complain about having two different modes, edit mode and retopology mode ...
there is an intermediate solution, some tools where possible, could be shared in the context menu ...
(I didn't read all the comments, maybe it could already be suggested)

@Pablo Dobarro (pablodp606) If we want complex semi automatic retopology tools, brushes and so on, we will want to have a new internal API that makes the development of these tools as easy as possible.

I am with Pablo on this, we should look up to the future and what might bring, retopo shouldn't be restricted to just mesh modeling tools..etc.

I agree on semi-automatic (also depends on what we mean with semi-automatic) and automatic retopology. That even makes sense as a new object type or whatever else might be the best way of doing it for both users and developers since the way of working with the topology could be so different.
Manual retopology might be best kept in Edit Mode.

From the developers point of view, I would first focus on the drawing mode and snapping to start with. These are the low hanging fruit but the most important ones, they don't need to be bound to retopology.

Then you can move to tools that can benefit edit mode even without snapping or retopo in mind, like strips and patches. To adress the point that it adds complexity to edit mode, that depends on the ambition you may have for retopology tools. If we get to some weird middleground in which we get both a Retopo mode and some of the same tools across both Edit and Retopo mode, that would be the worst case scenario for the developer. That's why it seems agree we should focus on just one mode, but not which mode.

The new tool system shines best when the tools do complex things with widgets which is what I hoped for, instead of just being slower ways of executing commands we already have. While in the past, during 2.7x, I would look at an addon or a potential feature and think "this would work better as a mode", now I think of the tool system to the solution to most of these. If it's really that hard to maintain the code of new tools, then why do we have the new tool system in edit mode in the first place? We'd have to go through this dilemma every time you have a semi interesting tool in mind, then. Hoping to not get locked out of a tool because it went to another mode.

Display
One of the major issues currently, is that we have no great built-in way to display your retopology mesh on top of a high res sculpt.

Technical note: this could be implemented by using z-buffer that contains only the backfacing polygons of the target mesh, so that front facing polygons in the target mesh do not occlude front facing polygons of the retopo mesh. Along with that you'd need to hide the backfacing polygons of the retopo mesh. If you'd want to show those backfacing polygons, they could be occluded by just the front-facing polygons of the target mesh.

YAFU (YAFU) added a subscriber: YAFU (YAFU).EditedFri, Aug 2, 11:44 AM

Is the recommendation to not to publish screenshots/videos of other softwares in design tasks to avoid legal problems still valid?

Is the recommendation to not to publish screenshots/videos of other softwares in design tasks to avoid legal problems still valid?

Yes. Please don't do this, we have a rule against it. The license of such software often does not allow it, I've removed such comments from this task now. Link to it instead, don't upload it here.

I think we need the edit mode to work perfectly for retopo as it is, with a few improvements, and a new mode for all the semi-automatic tools.
I don't see the fit for semi-automatic tools in edit mode, you need better snaping, a relax brush, a move brush, and few other, F2 give us a lot of usefulness.
But semi-automatic tools are a must, you can make a lot in really little time, and you can always go to edit mode to tweak some small imperfection.
If we have a new mode for retoplogy there is room to think new functionalities, and new retopo ideas that probably will never fit in edit mode.
And if we have a tab for retopology the edit mode in that tab can be already setup for retopology, so if you change mode in that tab from retopo mode to edit mode, you will have perfect setup for fixing small things, or if you want do all your retopo in edit mode.

Flo Tasser (TheFlow) added a comment.EditedFri, Aug 2, 7:13 PM

Have you guys seen this tool? It does a lot of what I think might be good to see shipping with Blender out of the box. Going to try it out for a few days!

Youtube: Blender 2.8 PolyQuilt addon
PolyQuilt BA.org Thread by the developer

we have a rule against it.

Ah, sorry man, I didn't now about that. 😉

Have you guys seen this tool? It does a lot of what I think might be good to see shipping with Blender out of the box. Going to try it out for a few days!
Youtube: Blender 2.8 PolyQuilt addon
PolyQuilt BA.org Thread by the developer

That's already a nice proof of concept right there.
https://twitter.com/sakanaya/status/1155444678226898945

The most logical solution to me is to create a new mode altogether like was originally suggested. The ideal retopology workflow would need things like specific brushes like a relax and smooth brush to fix tight geometry, which would muddy the waters regarding the differences between Sculpt and Edit Mode if they were added to the latter. The workflow would also need to automate things like snapping to either selected objects or every visible object, creating a new mesh over the original, and more. There are way too many steps needed before you can even start doing retopology right now, and if you start pushing all those tools into Edit Mode, you really can't automate many of these steps since it would mess up the current workflow for Edit Mode.

Retopology requires a specific type of workflow and a clean workspace with specific tools that are a mix of Edit Mode, Sculpt Mode, and even modifiers for maximum efficiency. If at any point during this workflow you have to switch modes and/or fiddle with modifiers to work with retopology, then you have compromised the efficiency of the design.

Also, pushing all the tools into Edit Mode would just bloat that mode even further. It's already hard enough finding new places to put shortcuts for easy access, and by mixing Edit Mode work with retopology work will only make this issue worse over time.

I am very much with @Wo!262 (wo262) and @Julian Perez (julperado) on the subject. For a manual retopology workflow a new editmode would be overkill. Especially if a new mode would mean that many of the edit tools will become inaccessible while the retopology shading is active. This could, despite all the good intentions, be even counterproductive. Focused tools are good for a workflow that is dedicated to a narrow set of tools. With manual retopology you often still need to switch back to regular modeling tools every now and so often as the lines beween standard modeling and snapping can be blurry at times. Manual retopo for hard surface can sometimes mean using the high-poly only as a reference but modeling the clean asset on top instead of snapping. Think of a bookshelf, for example. You wouldn't want to snap to the beveled edges here but rather have clean planes at first (and maybe bevel later).

Now all of this is of course from my current point of view:
Currently retopology is roughly 50% of my day to day fulltime work. I retopologize mainly CAD models for use in realtime applications or sometimes for animation. Over the years I have worked with ZBrush, Topogun2.0, Modo and Cinema4D (besides Blender). From these tools I find Topogun and Modo to be the closest with Modo being the winner for two reasons:

  1. I can access all the standard modeling tools in my regular modeling workflow and use them for retopology
  2. I can switch between models and topoview very fast in an already opened scene and that way retopologize a lot of small assets in one go (good for mobile games as well)

Dedicated topology tools are a brilliant idea. I am all for it but they may never compromise the existing tools. The standard modeling tools are just too important to be let go.
What Modo does for example is that retopology shading is just a viewport mode that can be turned on and off. Topology shading sets the active mesh to be shadedd in semi-transparent with backface culling in front of the other meshed which are (by default) shaded in a workbench-like shading mode. The dedicated topology tools are actually very few and consist of an automatic set of snap- and tool-settings (very much like the poly build brush in Blender). All the tools are tools which can be used in standard modeling as well but the specialized snapping retopology brush works best when used in the retopo shading.

I am actually taking a lot of the shading points already discussed on Blenderartists last year: https://blenderartists.org/t/blender-2-8-viewport-updates/1114082/412

What I would expect from a streamlined topology tool has actually already begun with the Polybuild Brush. For a topology tool it may need a few more additions:

  • Extrude a whole quad instead of only triangles (and then having to hold ctrl which interferes with snap toggle)
  • delete vertices, edges and polygons with a modifier key-click
  • move single vertices quickly with single-click/drag like with a sculpt brush
  • stitch vertices or edges together quickly (proximity merge)

The other things would already exist pretty conveiniently and fast. They maybe only need bettersnapping to geometry.

  • insert edge loop Ctrl-R
  • ring/loop select and dissolve (doubleclick and Ctrl-X)
  • fast bridging (loop tols - rightclick menue or own hotkey)

That is speaking of manual retopo of course.

In order of importance of the new features I'd say:

  1. What is really missing is the shading mode. That is the most important thing missing right now - with this even just the standard tools would be much easier to use on complex geometry, already.
  2. Next would be improvements in the polybuild brush to access the functions mentioned above in a way without switching between tools too often
  3. The more advanced thing then would be a new snapping mode. Here a normal based snapping would be really cool sort of like this:

Vector-based snapping mode - This works with any modeling tool from any angle: https://i.imgur.com/P0BVibV.gif
The same snapping mode paired with shring-fatten: https://i.imgur.com/g5MmxDS.gifv
Resnap selection to closes faces could also be a nice thing for this and could maybe also alleviate if loop-cut snap slide gives too much trouble being edited in realtime.
Those examples are also from modo, again.

Jut from my perspective what could come further down the line then:

  • the topo mode also displays precisely how many polygons the mesh currently has, how many of which are quads, how many are n-gons Maybe even lets the user input a target polycount and shows a small sort of health-bar indicating how many polys are left until the desired goal.
  • Resnap selection to closes faces would also be a nice thing and could alleviate if loop-cut snap slide gives too much trouble being edited in realtime. I imagine this something like: select a few faces, execute 'snap to closest' and all the vertices are projected to the closest surface like with the shrinkwrap modifier or when you do a 'proiject to subtools' in ZBrush.

What A dedicated viewport design looks and feels like can be tested right now by checking out ReTopoflow 2.0. I love what the creators of this tool accomplished as it has a lot of brilliant things going for it but the separate retopology mode makes me still ditch the tool most of the time, actually. When I do manual retopology - especially technical and hard-surface then the tools mostly specialized in sculpted character models sometimes fall flat. So I have to get out of the mode, struggle with the viewport settings just like if I had done the setup traditionally alltogether and go back into the dedicated mode again.
If a dedicated toolset ultimately is what is going to be decided on I would still urge the developers to please also make the shading mode available in the standard edit mode or pull over all the standard modeling tools into retopo mode as well. I can't stress this enough. Switching modes can create an immediate editing slowdown or papercut again if the tools aren't selected carefully enough.

Now, for semi-automated and fully automated retopology that's a different thing. Here I can see the user actually needing to use a completely different set of tools because they are guiding the algorithm. That's a dedicated logic like sculpting the way I would see it. Here the regular tools only come in in case you need to manually alter or edit the already finished auto-retopology.
I don't have too much experience with automated topo tools outside of ZBrush, though. So I will refrain from stating my oppinion on that topic.

The YouTube channel FlippedNormals just released a video tutorial on Blender 2.8 retopology. The video showed what they think is the most efficient way to do retopology right now without add-ons, where they discussed the pros and cons of the current tools that can be used for retopology. Since they are professionals and teachers of other 3D software like Maya and ZBrush, their input could be of some value in this discussion.

Link to video: Retopology for Beginners in Blender 2.8 - Retopo the Correct Way

Seeing this video and being reminded of all the necessary steps again I am starting to wonder if a dedicated topology mode could actually work. Now maybe this was the intention all along and I misread or misunderstood it but:
What if a dedicated topo editmode could be like an editing and snapping workbench, where the snap and view settings are turned on and are easily adjustable to do all kinds of topology work. It can also have additional tools if they really are tools which won't work in any other context than topology. Tools like guides for auto topology, guides for bridging rings, painted topology like ZBrush's TopoBrush which builds polygons where lines intersect. The regular modeling tools should still be accessible, though.

Also for what it's worth I can attest that these guys really know what they are talking about.
Thanks for the Link!

I have objections on making a completely new mode for retopology, there are simpler ways of enabling better retopology without bloating blender with highly specialized features that are not really useful for other tasks.

I would suggest:

  • X-ray/ display problem:
    • solution: Add a new display setting (Viewport Z-Offset) allowing the user to choose an artificial offset for the object. I think its simple for @Clément Foucault (fclem).
  • Snapping :
    • solution: new snapping mode. "Nearest object's face", not being screen space is not a problem, I think it's actually better to have it as global space, also add a collection filter to it.
  • Tools:
    • Tweak brush:
      • Proportional edit does just fine.
    • Quad strips:
      • Improve Ctrl+Click Extrusion to support drag events, and have it extrude multiple times while the user drags the mouse.
    • Quad Patches:
      • Make grid fill work in open loops and automatically correct the number of verts to produce a regular grid. *

May I also add: moving vertices without double click or pressing a key. Yes this can be achieved with the sculpt brush but this means switching an editing mode again.
Single vertices need to be moved often and fast to space them out evenly across the surface - especially in the beginning.

Ok, after discussing this more with @Pablo Dobarro (pablodp606), we agree to change the approach and try and incorporate better retopology support inside Edit mode.

This is how we think it can be done:

Display

Add a Retopology overlay display option, replacing the current Hidden Wire Edit mode display

Snapping

Add a Retopology snapping option, which allows snapping to self while also projecting onto source, and snaps the current selection.

Tools

We can improve certain tools, such as Poly Build, and also add more edit mode tools. Even though they may be more constrained inside Edit Mode, we can try to use this as a test-case for more advanced gizmos and tools in general

The original idea to have separate retopology mode is ultimately a consequence of how messy and overcomplicated current display and snapping settings are. What should be done is mainly a cleanup and redesign of snapping tools so that it's easier and more convenient to switch between them as well as using multiple modes at once. Rather than avoiding excessive complexity by separating whole retopo workflow into new, separate mode, the excessive complexity should be avoided by cleaning up existing tools to make them more universal

@Ludvik Koutny (rawalanche) it's partly that, and also that we wanted to add more advanced tools which wouldn't be easy to add to Edit Mode as it exists today.

Add a Retopology snapping option, which allows snapping to self while also projecting onto source, and snaps the current selection.

The key is, how are you gona define the "source" should it be the active collection, visible objects, selected objects or what?
Also, should the snapping be based on euclidian global space or projected screen space?

We can improve certain tools, such as Poly Build, and also add more edit mode tools. Even though they may be more constrained inside Edit Mode, we can try to use this as a test-case for more advanced gizmos and tools in general

I still think we should focus on improving current tools like grid fill and Ctrl+click extrude to support better retopo workflows rather than adding more and more specific tools that the user has to remember. Also bezier handles are pretty but unnecessary, proportional edit is way better imho.

I think the key is making those settings more clear reachable before tinkering about whether to add more tools or not.

@Ludvik Koutny (rawalanche) it's partly that, and also that we wanted to add more advanced tools which wouldn't be easy to add to Edit Mode as it exists today.

It seems actually easy to make polybuild a powerful tool for retopo, its already there and just needs a few teaks.

Yep. The general idea here is that it would be very sad if there was a "retopo mode" which would have some great tools useful also for regular poly modeling/creation, but users would have to constantly juggle between two different modes to access the good tools from both retopo and edit mode. Retopo modeling tools, if done right, are equally as useful in regular modeling of organic stuff, such as humans, animals, creatures or plants.