Wes Burke writes:
Jonathan Williamson is back with a brand new course on Retopology. As modern workflows have incorporated more sculpting, the previous techniques of box-modeling and poly-modeling have seen less use. In place of these older workflows is a new synergy between artistically-driven sculpting and technically-minded poly-modeling, through a process known as Retopology.
14 Comments
Easy retopology with Blender? Haha good one. There is Zbrush, 3D Coat or Topogun for this.
after just watching some tutorials for those programs to see how retopo is done in them i really don't see how it's different or easier than the method they teach for blender.
Watch some tutorials how to paint in MS Paint and Photoshop. Same right? Pick a color and click on canvas - totally same. But the rest is not. Retopo in Blender is tedious even with plugins like Retopoflow or BSurface.
Nah, ms paint and photoshop are not the same.
as long as we're using that comparison, from what i've seen your quick and easy retopology tools in those programs would be MS Paint vs the actual in depth methods and tools they (and blender) seem to share which would be photoshop.
there's a reason that in several of the tutorials i watched the people in them preferred or needed more accurate means that gave them good edge flow that followed the forms rather than what they were actually getting from the easy quick tools provided.
from some of the weird stuff i saw with zremesher and 3d coat do, i can't say i would want to use either for anything that really needed much more than just a quick remesh to improve polycount after a sculpt rather than an an actual finished retopology.
sure being able to tell it to follow certain lines here and there to generally direct edge flow may give you better edge flow than a mesh that is calculated without any input on that at all but it still doesn't actually mean you're getting good topology rather than just somewhat directed topology.
so honestly the programs you're putting up on a pedestal are only better for the easy things and even then not really much better than a remesh or decimate modifier since static meshes don't actually need good edge flow rather than good form/silhouette.
the reason that retopo in blender even with retopoflow is more tedious than your wonderful programs is because retopoflow was built for actual retopology work where it matters. which is tedious no matter what program you do it in.
just so you know the course this post was about didn't even use retopoflow or bsurface. the only addon it really used was F2.
I am not just addressing the quick and easy tools like zremesher. I am saying it's the whole ecosystem of retopology tools that these apps feature, the fact it is very well multi-threaded and actually performs on high density meshes which is not Blender's case.
Manual retopo is a must and it is tedious, but with Blender it is super-tedious. The gap is as big as between Photoshop and MS Paint, no kidding.
Yep I know they were using just F2, I've seen the tut.
mind giving me an example of this whole ecosystem of tools vs blender? not trying to be a butt or anything but i've been watching and looking for tutorials and such about all these programs since the start of the conversation and i really haven't seen anything that i thought was so much better than doing it in blender or really better at all.
i'm either missing something, just find it so much less impressive than you, or just have an easier time doing retopo in blender than most people. (honestly, i'm not the best at it but i don't find it all that tedious or hard so much as time consuming)
The end retopo won't be done better, same result can be achieved with Blender, the difference is speed.
These apps handle high res meshes a lot better. There are no performance issues with snapping to the surface, undo performance, viewport fps, and everything. This has a lot of impact on workflow speed.
The second thing is you have to take care of less things. The quads and edges will be connected automatically and the way it happens is often controlled by the order you place vertices. This has the second most impact on speed.
The third is the ability to use many automatic retopo tools that lets you establish the base, that you then manually refine where it is needed. There are some free apps that can also do this, but they are slower, not integrated, and produce uglier meshes that require more manual fixing.
Then there is not that important other stuff, that affects the ease of use which is also important. For example the viewport will render the backfacing geometry behind the high-res nicely or the UI is better designed and faster to use.
The ecosystem of tools I mentioned: I want to retopo 2 streams of simulated water into one mesh. I want to join them into one union mesh and automatically retopo them, because manual work would be time-consuming. Those meshes are non-manifold and Blender's boolean tools will badly fail after 15mins of computing on single core. Even if that worked, the shrink-wrap, decimate or remesh modifiers would produce a bad result. Zbrush or 3D Coat let's me do volumetric booleans on non-manifold meshes in seconds and then auto-retopo it with great results. What would have been several hour job is done in minutes. There are tools to re-project the detail on retopologized mesh from the original. In Blender baking displacement and throwing it into displace modifier on the lowres mesh with subsurf will produce incorrect (not the same) result as the original - especially on sharp edges.
I could write more use-cases and more points, but I think I made Blender incompetent enough. Topogun is $100. Even a student's or hobbyist's time is more expensive than that. And if you get Zbrush for $800, suddently a big door full of job opportunities and big money will open, so it's more an investment.
ah, so it is just that i find it much less impressive than you.
it is true that those programs handle higher polycounts better than blender though i've never seen a mesh that actually needed that high of a polycount for it's detail level.
i always decimate the high poly object before i retopo but even before the decimate i've tested with 4 million faces and i don't experience any issues or slowness with surface snapping from my retopo object to the high poly object with or without a shrinkwrap modifier. nor do i have any issues with fps unless i go into edit mode to actually edit the high poly object that way. why would i be doing that rather than sculpting? edit mode on my low poly retopo object still works well even with a high poly object in the scene.
granted mileage will vary based on hardware but that's just another reason to decimate the highpoly before starting the retopo. even on my previous budget level computer from like 2012 those issues are really only issues before the decimate and only in edit mode for the high poly.
as far as the second thing, that could be an advantage over an addon free blender but not really one against a combination of retopoflow / f2 / extruding + shrinkwrap / decimate / etc. it's really all about learning to use what you've got really. F2 speeds things along quite nicely if you bother to learn how to use it. very much like your stated automatic connected quads or edges. retopoflow makes simple parts much quicker, blender also has an auto merge feature, go into the options tab of the tool shelf and turn up the double threshold setting a bit so you don't have to manually connect/merge verts rather than just placing them close to each other.
third thing, personally i don't like refining topology rather than just making the parts i'm likely to have to refine manually and using retopoflow for quickly generating the easy bits and i don't even have to manually connect / merge the two because of the aforementioned auto merge feature.
as far as the other stuff you mentioned. that sounds a lot like turning on blender's x-ray feature though i'm either not sure what you mean by backfacing or you either mean something different than how blender uses the term or don't know why you'd be trying to mess with the retopo on the other side of the mesh that you aren't looking at... but they did mention using blender's x-ray setting and the hidden wire settings for easily seeing your retopo object wires over your high poly. you sure you actually watched the tutorial series?
UI design is opinion which you're welcome to have one different than mine. though i fail to see how it's going to be faster to use than a hotkey based workflow like blender's where the visible ui doesn't factor in much for activating commands. their ui maybe easier to learn but that doesn't mean easier to use.
there's a reason blender gained a reputation for being great for quickly modeling things. if you know what you're doing retopo is basically modeling with a few small changes and a bit of set up so that you don't have to push your verts around as much as when you're actually modeling from scratch.
that's a very specific use case with the fluid sims. i've never had to do that for any reason... not that i'm a large sample group. that said it sounds like most of the triumph there isn't anything to do with being better at retopo rather than having better boolean tools... and if your purpose was to join the two streams into one mesh wouldn't it have been accomplished by the boolean operation? in which case why would the decimate or remesh modifiers not give a good result?
blender's fluid sim creates single meshes anyway... rather than slapping two sims that probably weren't interacting properly together it'd be better to just make a new one with multiple inflows.
also, why the heck would anyone even think to create a fluid sim then destroy the detail by turning it into a low res mesh in order to try to add that back in with a displacement and subsurf? if you're adding a subsurf to get back detail you just tossed away why bother killing the detail by making the dang thing low res to start with?
the heck kind of stupid blender use case are you trying to push with that? just make it high poly, use decimate and tweak the settings until you get your own nice trade off of detail vs poly count.
please though, don't stop with just the one use case and do give more points.
Have you used 3D-Coat or topogun extensively? If you do alot of retopo then id suggest you give them a try. You'll be presently suprised how easy retopo can be.
No ones saying blender cant do the job, people are saying that it could stand to improve massively. .. which all programs could ;)
nah i've just been watching a crap ton of retopo tutorials for them and zbrush since the start of the conversation. in all honesty i haven't seen anything in them that hasn't really been covered by a combination of retopoflow, f2, and what blender already has.
i stated before i don't actually find retopology in blender tedious.
i'm not saying blender retopo is great by default. that's obviously not the case. i'm just saying that there's not any reason to think it can't be done easily in blender.
I see that you are still using blender 2.6 or below ;)
Funny. And I see that you know very little about the world outside Blender.
Blenders Retopo tools are quite a bit behind alright. 3D Coat is my goto.
i gotta say CGCookie is worth the money. not only are they good with art but they're also good at teaching it.