Jump to content
3DCoat Forums

ZBrush 4R6 - Yet not excited


Recommended Posts

  • Advanced Member

Funny thing...I used to bananas when I got a new version of ZBrush and while I'm happy they're working to improve it more, the announcement of 4R6 with ZRemesher didn't really hit home with me. I use 3D Coat for all of my retopo work and don't see ZBrush being a better alternative. It's OK in a pinch, but 3D Coat has had most of this functionality for a while, as far as I'm concerned. I would say that Autotopo is a tool I almost never use now, but I do use strokes pretty often. In ZBrush I had experience with QRemesher, Decimator, etc., but I'm just not all that excited. I wish they had done more with projection, shaders, the auto-updater, and paint layers in ZBrush. Even the auto-updater is a joke, as all it does is install the new version in a new directory and copy certain files from the previous install. This takes up TONS of room on the drive and is very archaic. I still use both apps, but I've quickly become a die hard 3D Coat user.

Edited by alvordr
Link to comment
Share on other sites

  • Advanced Member

Before I posted, I downloaded the new ZB update and tried it out. I ended up with weird edge flows and some other odd artifacts on different models. I wasn't impressed. I wouldn't personally compare Autotopo to anything in ZB, as I don't find it extremely useful, unless I don't care about edge flows and just need to export a sculpt with geometry for a quick render in another app. I almost always use the manual retopo in 3D Coat.

I don't believe that ZB is using decimation (although I could be wrong), as Decimator triangulates meshes, so unless they're doing that and then quadifying it, it seems unnecessary. I believe they're using an algorithm and guides to detect proper edge flow, but it doesn't seem near ready enough to be in an update.

The one thing I did like about the ZRemesher was that it was faster than QRemesher.

Edited by alvordr
Link to comment
Share on other sites

  • Advanced Member

Did a quick demo of my own, no guides on any app, all default settings, except for driving the polys to 1k.

ZBrush 4R6: ZRemesher yielded 1787 polys

ZBrush 4R5: QRemesher yielded 1072 polys

3D Coat: Autotopo yielded 1007 polys

Screenshots below. You can see that ZRemesher did the best with the flow in the palm, but horribly with the fingers (keeping in mind no guides on any of these). 3D Coat did better on the fingers (but about the same as QRemesher) and managed to hit the 1K mark more accurately. QRemesher was not bad either... None of the resulting edge flows (without guides, mind you) were great.

Of course, I'd rather do this by hand (ehem).

ZRemesher:

zremesher.jpg

QRemesher:

qremesher.jpg

Autotopo:

autotopo_vs_zb.jpg

Edited by alvordr
Link to comment
Share on other sites

  • Advanced Member

One other thing I noticed is that QRemesher in 4R5 couldn't go below 1k, whereas ZRemesher in 4R6 could and works much faster. That's nice. However, even if I told it to take that hand model and go to .25 k (or 250 polys), it put it at 916 polys. That becomes a lot of guesswork to get the right poly count, and I wonder if it matters which model you're using in determining how accurate it will be.

Edited by alvordr
Link to comment
Share on other sites

  • Advanced Member

@alvordr

For hands, please use guides.

About Zremesher, please read the manual first. It adds some more loops so you have 1700 faces. You can avoid it. (try alt, the optional algorithm too)

About the 3dc autopo, you have some impossible to edit topology there, you also have 7 n-gons. A rather chaotic palm.

Well, I have some excellent results using - testing z remesher. The good news? I can easily edit these meshes. Rather impossible for 3dc or qremesher cages.

Guides are needed to avoid spirals on hands -fingers legs... the usual nightmare. LOL

but you tested it on hands man... hey. You need one hand - base only. If you work on a multi resolution application I mean. Right? ZBrush is one of them.

  • Like 1
Link to comment
Share on other sites

  • Advanced Member

@alvordr

For hands, please use guides.

About Zremesher, please read the manual first. It adds some more loops so you have 1700 faces. You can avoid it. (try alt, the optional algorithm too)

About the 3dc autopo, you have some impossible to edit topology there, you also have 7 n-gons. A rather chaotic palm.

Well, I have some excellent results using - testing z remesher. The good news? I can easily edit these meshes. Rather impossible for 3dc or qremesher cages.

Guides are needed to avoid spirals on hands -fingers legs... the usual nightmare. LOL

but you tested it on hands man... hey. You need one hand - base only. If you work on a multi resolution application I mean. Right? ZBrush is one of them.

Michalis,

I know to use guides, but this was about the test. It was taking the least time-consuming approach and shows how well the software will operate where/when guides aren't used. Without guides, the fingers were better retopo'd both with QRemesher and 3DC (no removal of edge loops needed). As far as your comment about 3dc meshes to be easily edited, this is simply not true. Perhaps it's a matter of how well you know the software. We can debate this all day, but I understand that your experiences are going to be different from mine. The test was actually done on "one hand" as you stated, not hands. Like I said, I prefer to retopo manually, regardless. None of these solutions are useful for me, except when just simply getting a quick render in an app that doesn't support Voxels.

Edited by alvordr
Link to comment
Share on other sites

  • Reputable Contributor

So Andrew needs to update the math in autoretopo.

This is what it boils down to. I think this is where it pays off having extra development help. Andrew has been so busy improving other areas of 3D Coat, he has not had the time to go in (as it appears Pixologic has done with this 4R6) and refine the algorithm. I think they added a surface curvature analysis to the edge detection algorithm, and then instead of the user having to spend time manually cleaning a dense mesh up, they added a subD reversion feature. 3D Coat does not have this.

I asked Andrew over a year or so ago, if he could add something like this or contact this guy, who wrote a cheap plugin/script for 3ds Max, to reverse subdivision. He never answered. Sometime AutoPo leaves you with too dense of a mesh. If you redo with fewer polygons, sometimes the result is much worse. This leaves the user with a lot of cleanup work to do or scrap it and do the whole thing manually. The latter is what I've had to do most of the time. Where I think this SubD reversion routine would be most helpful of all, and this is how I expressed it to Andrew....is that exporting from 3D Coat with the "Dense Quads" option would be greatly enhanced with SubD reduction/reversion capability.

What I think would be great for refining the Auto Retopo algorithm would be a departure from using evenly spaced quads/edgeloops. Perhaps, using surface curvature analysis, apply a decimation routine (fewer polys/edgeloops where there is a relatively flat surface, and more where there is a higher degree of incidence angle) to his dense quads routine.

Another option might be to utilize Vertex color to allow the user to paint-select with white (CTRL + paint-select), areas where there needs to be much less density. Also, since we have the ability to place points on a voxel object, with the FIT and PLANE tool, add an option (probably for more advanced users) to place points where you want your 5 point intersections (usually where 2 different edgeflows merge. I'm thinking this might help the algorithm "direct traffic" by placing a stop sign, as it were, and saying this is where you should terminate these two with a simple 5 point intersection....instead of a convoluted group of ngons or starting a spiral edgeloop.

Link to comment
Share on other sites

i was reading this:

http://www.polycount.com/forum/showpost.php?p=1868921&postcount=60

Pixologic put Tomas Pettersson (the Sculptris creator) to develop a better algorithm

--------------------------------------

Surely could Andrew put effort to refine autoretopo.

But for now i find far more useful to put all the dev time refining the Paint Room.

Improve the paint room -for my point of view- must be the community/betatesters focus.

Link to comment
Share on other sites

  • Reputable Contributor

By the way...I don't mind giving kudos to Pixologic for enhancing this toolset, or "one-upping" Andrew in this regard. It shows that this sort of refinement is indeed possible, and it makes the whole toolset far more useful. I hope Andrew takes notice. Competition is good for both userbases. Think about it for a moment. Had Andrew not introduced this Auto-Retopo toolset, it would not exist in either MB or ZB either. They get to benefit from Andrew's innovation. What bothers me is when their userbase gives them the credit for "innovating." The refinement is innovative...yes. Z Remesher in general and their new adoption of the paint selection for dense areas, is not. I wish their userbase would get it straight. They go ga-ga over this stuff and for the most part, Pixologic, in the whole R4 cycle has just cherry picked unique ideas/innovative tools they saw in 3D Coat.

I don't blame Pixologic for giving their users what they ask for, as it can work both ways. But it does seem a little odd to see such a large company resorting to this kind of practice on a wholesale level. Usually it's the other way around....it's normally the new kid on the block, or smaller companies doing this in attempt to catch up with the bigger ones. I honestly feel like the folks at Pixologic take 3D Coat far more serious than most of their userbase. They've noticed that Andrew has closed the gap, within just a few short years, from being just a toy of sorts, to a very serious competitor...to the extent that it's close to having replaced Mudbox as it's nearest competitor.

I just hope that Andrew returns the favor takes some pages from their playbook, as it were.

  • Like 2
Link to comment
Share on other sites

  • Advanced Member

This is the sort of discussion I was hoping to get out of starting this thread. It's important to recognize the merits of development in all the apps we consider using, but at the same time constructively looking at the actual usefulness. AbnRanger is right, in that I see a lot of ZB users going gaga over stuff before they've played with it (which I used to do, as well). I'm not trying to belittle Pixologic's work and if it's one thing I'm gaga over all the time with ZB is that I don't have to pay to upgrade. If they stopped right now, I would be super happy with ZB in it's present state. The features it lacks I find in 3D Coat and realize that most of what I do and need are already in 3D Coat.

Link to comment
Share on other sites

wise words AbnRanger, let me say

Pixologic have 25 members in the dev team, and one man labor slowly is mining their supremacy.

Its ZB a standard app ? sure

I think is better to make 3DC more strong where is strong, first.

And be competitor in a clear empty market zones

- Be an alternative to professionals that cant afford Mari, taking care of standards: LUT, space color, gamma correction, create layers by channels, folder groups, tiles over using multiple UV maps... etc

- Be an alternative to Substance Designer adding procedural painting.

- Add sculpt layers

This -may be- can put Mudbox in 3rd place

regards

Link to comment
Share on other sites

  • Reputable Contributor

wise words AbnRanger, let me say

Pixologic have 25 members in the dev team, and one man labor slowly is mining their supremacy.

Its ZB a standard app ? sure

I think is better to make 3DC more strong where is strong, first.

And be competitor in a clear empty market zones

- Be an alternative to professionals that cant afford Mari, taking care of standards: LUT, space color, gamma correction, create layers by channels, folder groups...etc

- Be an alternative to Substance Designer adding procedural painting.

- Add sculpt layers

This -may be- can put Mudbox in 3rd place

regards

There already is a good deal of procedural texturing available in the Paint Room, via the FILL tool. That's what makes it so darn powerful. Remember, you can freeze parts of your texture maps in order to isolate the areas you want the fill to be applied to. You can even use an image map as the procedural texture (you can see a little glimpse of that in the V4 promo video)....that aspect is so cool. But it can be improved. I know a very talented programmer here in this part of the US that loves Substance Designer, and maybe I can coax him into developing a similar plugin/script for 3D Coat, to expand that functionality a good deal. He said he bought a seat of 3D Coat, some time ago...so maybe we can get our first 3D Coat plugin! :D

Nevertheless, I really do wish Andrew would expand procedural capability in both the Paint Room and in the Voxel Room. The General Brush (LiveClay toolset) has a few noise patterns and it is so handy in the right situation. I asked him and Raul to expand it to Reptile patterns, animal skin patterns, wood patterns etc. I think this is one of the key features in ZB (procedural noise library?).

Getting back to ZB 4R6 and the refined Auto Retopo algorithm...I am very encouraged to see this, actually. I don't think Andrew will sit idly by and leave Auto Retopo as it is. When they copied the Sketch tool and improved upon it, he reciprocated and made the Sketch tool better.

Link to comment
Share on other sites

But for now i find far more useful to put all the dev time refining the Paint Room.

Improve the paint room -for my point of view- must be the community/betatesters focus.

Hopefully Andrew going into stealth mode lately has something to do with that. The paint room needs a ton of work, especially considering it's the most likely reason for ZBrush owners to try/buy 3DC atp.

  • Like 1
Link to comment
Share on other sites

http://www.zbrushcentral.com/showthread.php?96408-Playing-with-polygons-topology

DrPetter

Today I'm turning his entire polygonal skin into quadrilaterals, using an automatic algorithm that I am currently developing.

When Alpha 6 has been released, I will continue working on a bunch of exciting new features like those discussed in this forum thread.

-------------------------------------

and more:

http://meshmixer.com/forum/index.php?topic=1577.msg3087#msg3087

http://forums.luxology.com/post.aspx?f=4&t=62181&p=560194

http://www.polycount.com/forum/showpost.php?p=1868921&postcount=60

Link to comment
Share on other sites

  • Advanced Member

retopo%20compair2.jpg

looks like we have a winner here. zbrush even gets those spikey details near he mouth.

one thing that 3dcoat produces is those spiral topology near the head which is why I don't use autoretopo much in 3dc.

Link to comment
Share on other sites

  • Member

One other thing I noticed is that QRemesher in 4R5 couldn't go below 1k, whereas ZRemesher in 4R6 could and works much faster. That's nice. However, even if I told it to take that hand model and go to .25 k (or 250 polys), it put it at 916 polys. That becomes a lot of guesswork to get the right poly count, and I wonder if it matters which model you're using in determining how accurate it will be.

Did a quick demo of my own, no guides on any app, all default settings, except for driving the polys to 1k.

ZBrush 4R6: ZRemesher yielded 1787 polys

ZBrush 4R5: QRemesher yielded 1072 polys

3D Coat: Autotopo yielded 1007 polys

Screenshots below. You can see that ZRemesher did the best with the flow in the palm, but horribly with the fingers (keeping in mind no guides on any of these). 3D Coat did better on the fingers (but about the same as QRemesher) and managed to hit the 1K mark more accurately. QRemesher was not bad either... None of the resulting edge flows (without guides, mind you) were great.

Of course, I'd rather do this by hand (ehem).

ZRemesher:

zremesher.jpg

QRemesher:

qremesher.jpg

Autotopo:

autotopo_vs_zb.jpg

Did a quick demo of my own, no guides on any app, all default settings, except for driving the polys to 1k.

ZBrush 4R6: ZRemesher yielded 1787 polys

ZBrush 4R5: QRemesher yielded 1072 polys

3D Coat: Autotopo yielded 1007 polys

Screenshots below. You can see that ZRemesher did the best with the flow in the palm, but horribly with the fingers (keeping in mind no guides on any of these). 3D Coat did better on the fingers (but about the same as QRemesher) and managed to hit the 1K mark more accurately. QRemesher was not bad either... None of the resulting edge flows (without guides, mind you) were great.

Of course, I'd rather do this by hand (ehem).

ZRemesher:

zremesher.jpg

QRemesher:

qremesher.jpg

Autotopo:

autotopo_vs_zb.jpg

One other thing I noticed is that QRemesher in 4R5 couldn't go below 1k, whereas ZRemesher in 4R6 could and works much faster. That's nice. However, even if I told it to take that hand model and go to .25 k (or 250 polys), it put it at 916 polys. That becomes a lot of guesswork to get the right poly count, and I wonder if it matters which model you're using in determining how accurate it will be.

This is a really cheesy work around and requires minimal re-sculpting in ZB, but if you are looking for edge loops for animation with the joints of the fingers and what not... you can snap a few thin tube bracelets around the joints... that will give enough difference for true edge loops to be formed where you need them... as such it's best to do the zremeshing before high detail. As it is now zremesher tends to create spirals without some finessing.

Link to comment
Share on other sites

  • Advanced Member

This is a really cheesy work around and requires minimal re-sculpting in ZB, but if you are looking for edge loops for animation with the joints of the fingers and what not... you can snap a few thin tube bracelets around the joints... that will give enough difference for true edge loops to be formed where you need them... as such it's best to do the zremeshing before high detail. As it is now zremesher tends to create spirals without some finessing.

I love how I often get the best ZBrush tips on the 3DCoat forum :-)

Link to comment
Share on other sites

  • Advanced Member

No spirals in this case. I used bracelets - guides.

Just a 5 mins doodling (dynamesh), z remesher, exported in blender. UVs (follow active quad mode) and a ready made bump + an already baked AO map).

Took ~30 mins including rendering. Ready to export. I can even delete half of these loops but I don't mind.

I think, z remesher is an amazing tool. IMO, the first really working auto retopology.

post-2454-0-25662700-1372580532_thumb.jp

post-2454-0-36654900-1372580549_thumb.jp

  • Like 1
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

 Share

×
×
  • Create New...