3D Coat's Brushes
Posted 30 March 2012 - 01:19 AM
I have recently given colleagues (avid Mudbox/ZBrush users) a brief overview of 3D Coat and whilst their comments were positive overall (mainly referring to its "potential"), the brushes really seemed to be problematic for them. The feel and general performance wasn't something they were particularly impressed by. Larger brushes tend to feel sluggish and weird artifacts often show up - even after adjusting the spacing settings, interpolation, brush alphas used and so on. Another thing was the difference between surface mode and voxel mode - there doesn't appear to be much in the way of consistency between the way the various brushes feel and perform. Some people seemed to prefer using the Grow brush as a standard brush when sculpting with voxels whilst others preferred Airbrush - the one common thing was that neither felt ideal.
As for artifacts - with each stroke, various brushes seem to start and end with a mound/dimple. Playing with the spacing settings seemed to affect this behaviour but should *any* brush really be doing this (especially default brush behaviour)? This side of 3D Coat doesn't seem as robust as it should be and I'd personally rather have fewer default brushes that work flawlessly rather than a whole bunch that all seem to have problems.
I'd be happy to post some examples of brushes that don't perform particularly well but I'd like to hear whether anyone else has shared similar experiences or feels the same. Perhaps it'd be useful to use this thread as a place where we can experiment with various brush settings and try to pinpoint things that aren't right. I'm sure that if we can help Andrew identify the issues then he'd be all the more willing to address them.
Posted 30 March 2012 - 06:06 AM
Some of the regular surface brushes with remove stretching enabled cause artifacts. I submitted a bug report a couple of months ago about it. I do not seem to have the dimple problem you mentioned. The linux version did have the dimple problem but it was squashed a few versions back.
Another consideration is the voxel resolution in voxel mode and the polygon count in surface mode and how that effects the brush behaviour...
As mentioned above 3DCoat's digital Clay handles differently plus Voxel and Surface mode are different by design. The brush behavior will be different.
I agree that voxel brushes could use a going over, Surface mode and LC brushes I feel I have more control over the outcome.
I am not a professionally trained artist so it would be interesting to see how some of the more experienced users tackle brushes in 3DCoat, of course I will put
in my 2 cents worth...
Posted 30 March 2012 - 06:54 AM
Voxel and Surface mode are different by design. The brush behavior will be different.
The subtleties of which confuse the new user. The penny might drop more if users were dropped into surface mode first.
What happens is that the first time user ( or in reality user of other apps ) has already experienced a raft of tools that do not behave to their expectation derived from other applications tools.
For example the difference between surface modes clay and standards modes clay is staggering. Surface mode clay is evidently superior yet the first time user is exposed the the more peculiar standard mode clay. Surface mode brushes do fulfill first time users expectation to a greater degree. At the moment you have to swap between two rooms to find the best brush for the job. John mentions the airbrush and grow and quite rightly - yet surface mode has no performance equivalent as yet.
The four brushes that John and many others mention should in actuality be the first top four in the toolbar of both rooms - the rest divided. Those four tools should then be easily interface swappable for the new user who prefers other favorite tools as experience allows.
Every user in ZBrush has their preferred top four brushes. What they are depends per person but each fulfill / answer a basic task orientated need and most are consensus or get the nod of respect from users as personal preference but nonetheless worthy and or equal.
In truth the work has already been blazed by zbrush in this respect - find out the four most commonly preferred brushes by task , finesse the brushes closest by task in 3dc and there's your answer for the first four brushes :-)
It's the first point of contact, the initial fulfillment of standard expectation from there the other depths of 3DC could reveal themselves. if a brush evidently behaves better with spacing on why isn't it's defaults setting for that tool set to spacing on?
The default e-panel setting is misleading - it should lead with brush pressure affects depth first - then draw with constant depth and radius regardless of brush pressure then thirdly the present default e-panel setting "brush pressure effects radius and depth" which is a better finishing setting but a lousy generic setting. Changing this would make an immediate and big difference.
Myself one of the things I absolutely adore about 3DC's brushes is the range of polishing modes hidden within all it's tools smoothing modes. They are exquisite and each a brush in their own right . This is hidden from the first time user John as they also perfect the ills of which you speak yet this is not evident at first. With time these smooth modes become such that sculpting in 3DC to my mind is more pleasurable for many tasks than ZBrush but initially more painful. The pleasure of 3DC with further investigation is truly finding all the things it does exceed ZBrush by way of intuitive sculpting and there are many. For this reason I will swap about for the present.
With regard potential , if a tool can shave time off my exposure to sitting at a computer with regard achieving certain tasks then in this sense it is already fulfilling potential . Sometimes a production assessment of potential can be clouded by an existing investment in habit.
Posted 30 March 2012 - 08:15 AM
That being said, I do realize that having brush tools (in addition to the other great tools) is necessary, and some people would like to simply stick with using brushes to build most of their sculpture. Therefore, I suggest that the 3D-Coat developers work on creating a brush which is identical to the Clay Buildup brush in Zbrush. That brush is the best brush I have ever encountered in any sculpting application. I have tweaked some brushes in 3D-Coat to be similar to the Zbrush Clay Buildup brush, but they are not quite as good. Also, a brush like the Knife brush in Mudbox would be nice. I have tried using the Pinch, CreaseClay, or Grow brush with a pointy alpha, but a dedicated Knife brush would still be better.
Nvidia Quadro FX 3800 (1GB GPU memory, CUDA, OpenGL)
Windows Vista Professional 64-bit
1st display is Wacom Cintiq 21UX (2nd Gen) (1600 x 1200)
2nd display is ASUS VE245H (1920 x 1080)
Posted 30 March 2012 - 08:38 AM
As for artifacts - with each stroke, various brushes seem to start and end with a mound/dimple. Playing with the spacing settings seemed to affect this behaviour but should *any* brush really be doing this (especially default brush behaviour)?
Using the lines tool options in the E panel has this happening at the end of the stroke all the time, especially when intersecting. I would be very interested if anyone knows how to prevent this easily.
Win.7 64 . DirectX 11.0
Intel Core i7
NVIDIA GeForce GTX 680
4.0.04B1. CUDA 64
Posted 30 March 2012 - 12:59 PM
tis true .
I suggest that the 3D-Coat developers work on creating a brush which is identical to the Clay Buildup brush in Zbrush. That brush is the best brush I have ever encountered in any sculpting application.
great point Timmy.
Personally I think the greatest thing about 3D-Coat is being able to get so much good work done without using the brushes at all.
As a useful btw -to get a standard brush feel in Surface Mode > LiveClay>General Brush >check the airbrush mode>choose the small sharp alpha/brush >;set falloff to 30 and focal shift to 3 >cntrl click and make a copy of your settings and name it
......also there's another standard brush type in the surface mode brushes - strangely it's in the chisel tool with the cntrl key pressed. The Chisel tool is a gem - chisels with left click - soft builds with cntrl click and polishes through form with shift click. 3DCoat is full of gems like this.
Also there's another major difference between standard voxel mode and surface mode and something I make a point of repeating in some of my vimeo tutorials and that is that the surface mode sculpting brushes are by general rule planar in their nature like zbrush's claybuild up brushes whereas the standard mode brushes are more rounded in their nature and shape profile. For softer slower control Standard Mode is a great option - for faster roughing in choose Surface Mode.
The surface mode Move tool is like ZBrushes snakehook brush and move tool combined. It can both be used to draw and sculpt as well as smooth wonderfully. 70 -80% of what all the other sculpting tools can create can be created with the move tool alone but in Pilgway's typical modesty - no big deal is made of this but maybe should be.
Can you share LiveClay brushes yet? Being able to share brush tools may be a way to establish easier conversation with regard some useful generic artistic consensus on brush behavior - particularly so if e-panel setting would be saved with them.
Myself I really hope that Andrew allows us to have temporary swappable tools where we can be in our main tool of choice but can temporarily hot switch and cycle through a collection of tools of our choice .
Posted 31 March 2012 - 01:58 AM
Agreed. 3D Coat took a major step forward in this area back when Andrew did all the Multi-Threading work, then re-worked the Brush Engine. However, that didn't get 3D Coat to the finish line...just a closer step to it. It's still has some distance to go. I hope he makes it his mission to improve performance (and stability) until 3DC has closed ranks with the competition. It's not lacking for features, in my opinion.
I installed the newest one today and I can't say I saw any difference to speak of with the larger brushes. I was trying to paint a 4K map and even moderately large brushes just fell apart. They stop working at all, start jittering around, or just act very laggy. Same problem as always for me.
I was only painting the diffuse btw, no depth, no spec.
There is still a major need to improve brush performance for hi-res work, but at least 3DC is kick-ass for detail stuff. Small brushes and sharp detail are way better than Mudbox IMO
Posted 31 March 2012 - 10:45 PM
There are some really informative replies on here with some good tips too! We all know 3D Coat has some awesome features and we've rightly praised those areas where it excels, however, the general consensus does appear to be that the brushes need some work so I'd like to start off with a picture highlighting a very annoying problem:
For now, I've decided to stay away from more complex brushes as it'd be a good idea to absolutely nail the 'staple' brushes that any 3D sculpting software should have. The first example is taken straight from ZBrush - I've just used a simple mouse gesture on a sphere and the results are pretty self explanatory. I'm not sure why activating the 'Use Spacing' option removes the mounds/dimples at the start/end of each stroke but being as though that's the current default behaviour, I personally feel it really needs to be addressed. When sculpting, having to permanently smooth out the errors sucks away a lot of time and more importantly, breaks one's creative flow.
I've noticed that Use Spacing affects performance quite drastically and the higher the value, the more you begin to notice the mounds/dimples, which seems a little bizarre to me? It'd be nice to have a solution that doesn't mean having to take a performance hit. ZBrush in this regard comes out on top right now (especially on my computer).
Another thing that I sneakily put into the first ZBrush example is a single stroke made with the Dam_Standard - anyone that does character work will know its value for creating wrinkles or clothing etc. Yes, you can get close in 3D Coat but the point is, it takes a lot more work (smoothing off edges etc) to get the same result. I personally haven't been able to achieve a similar stroke in 3D Coat using voxels, surface mode, or any combination of strokes and alphas. For me, these brushes performing consistently form the absolute basic requirements for a sculpting package.
It'd be interesting to go into this in more detail as I feel the results and subsequent improvements would elevate 3D Coat substantially.
Posted 31 March 2012 - 11:15 PM
Posted 01 April 2012 - 01:32 AM
Now,the question is why is there a large range of difference in the stroke pattern between yours and mine. I do like this thread because if we can nail down what is causing problems then the brushes can be truly addressed.
Edit: The difference here on the most part is due to mine being a a low polygon sphere. This one is 250,000 polygons...
Posted 01 April 2012 - 02:01 AM
I seem not to get as bad of areas on the end of the stroke..
Wow, Border size makes a difference, thanks for that tip...
The second pictures is from increasing border size in the E-Panel.
Posted 01 April 2012 - 02:24 AM
These issues are what many first time 3D Coat users are experiencing when coming from either Mudbox or ZBrush so it isn't creating the best impression to those that know and have experienced better.
Just like to add that I hope Andrew doesn't feel overwhelmed by all of this. It shouldn't be viewed negatively and I think the reason why it's important to focus on these issues in such detail is because they're so important to the long term success of 3D Coat as a sculpting package. Good brushes will underpin everything! Also, it's really frustrating because it's so close to being right yet all these little problems are holding it back.
Posted 01 April 2012 - 02:36 AM
I learned to work around the brush dotting but I would think in a production schedule with deadlines this issue would be killer. Clean strokes and solid behavior would be needed for sculpting work...
I'm glad this thread was started not as a bashing arena but as a real concerned effort to help Andrew in his development of 3DCoat. I will be out of town but when I get back I will continue...
Posted 01 April 2012 - 02:44 AM
Posted 01 April 2012 - 03:15 AM
I figured ol' Frankenstein would be a good character to do some damage to with some brush experimentation so I have just created a rough base shape for us to use.
Grab it here: FRANKENSTEIN
Posted 01 April 2012 - 03:21 AM
We clearly all manage to get decent results out of 3D Coat which is why we all obviously use it. The issue seems to be that a lot of us have to adjust our painting style or spend ages configuring settings in order to achieve the results we're looking for. This obviously means that the default brushes aren't doing a great job and that there's a great deal of room for improvement - I think that's probably the key point we all need to focus on. I am happy to provide a test head or similar for us to test stuff out on - it'd probably good to be testing these brushes out on something with a reasonable poly count and with absolute targets we can try to create (wrinkles, hair, features, etc).
That would be great John to have a test model so we are all working and testing brushes and behavior on the same model...
I think you nail it for it for having to adjust brushes and create some workarounds to get the brush where we need it. A solid default brush would be better and then we could fine tune as needed.
Posted 01 April 2012 - 09:12 AM
To be honest for me 3dcoat is good mainly for concepts,you can start and define the shape at a level that after can be used in other software(let's say it's useful for a prototype),but if you need a final version the brushes(or the tecnology)don't allow to do it.
A part from the issue pointed out above I would like to be able to smooth the difference between height you can have when you start to do a complex height shape following your path stroke,it could be really useful for some detailing work,practically it's like when you do fur combing in some software and you smooth only the length of the fur and not it's direction(with 3dcoat the results are always bumpy).
Posted 02 April 2012 - 08:56 AM
I have myself been pushing for the addition of all spacings and e-panel settings being recorded with all brushes to set better defaults and thus reduce some of the findings you describe. There are at present irregular behaviors with settings that get saved or not with tools. My extrude settings in my latest version do not for example get saved between application uses , nor keep their spacing settings.
Of note between applications: take a tool like extrude or draw - the two closest to the polygon burying facility of the a dam standard in ZBrush - at different sizes they effect the surface differently . in ZBrush much of the effect is the same/similar at larger or smaller sizes but not so in 3DCoat. In 3DCoat brush size is everything and will dramatically alter your marks. Is this then a flexibility or weakness of the tool? The Airbrush tool depending on size can dramatically pinch/ sharpen edges or not - often a larger brush will pinch more than a small one.
Also of important note here: many of the standard mode tools set to e-panel strokes modes are realtime "additive" on new empty layers based on voxel information beneath them (much like the zbrush canvas response for tools but genuine 3d instead - this too is wholly different to the ZBrush approach and plays some part in what you describe. Thus for a like for like comparison , the discussion would be best simplified to surface mode responses solely.
To repeat, better performance is already an option if e-panel setting and spacing settings could be saved.
If this is added we all would benefit right now.
p.s: the subject of brush control a "repeat last stroke option" in 3DCoat would greatly assist control when building precisely over stroke paths.
p.p.s: Perhaps a community poll on development focus priorities may assist Andrew target his time based on user feedback? Where does brush development stand as a priority compared to other wishes?
Posted 02 April 2012 - 02:05 PM
I believe the defaults need to be sorted out but as demonstrated in my previous images (and your own examples), there are some issues with the brushes that need to be resolved as they interrupt and affect the quality of any sculpt within 3D Coat. The bumps and lumps we all often have to spend time smoothing out do appear to be visible in a lot of the organic sculpts that 3D Coat users create (not their fault either). The level of detail and quality regularly seen on ZBrush Central appears to be somewhat unattainable with 3D Coat's sculpting tools - if that quality is possible, then it probably requires a lot more time.
I've worked in the games industry for 14 years creating both concept art and 3D in-game art. I state that because I know that most of my colleagues, friends and former colleagues are completely aware of 3D Coat but all prefer ZBrush or Mudbox when it comes to creating details and finishing off a model. I know sculpting isn't necessarily the main attraction of 3D Coat and there's much more to it, but I believe it'd be worthwhile for Andrew to spend some time improving the brushes. Therefore, the point of this thread is to try and identify the weaknesses and provide some feedback from artists so that Andrew has a chance to make it right. As it stands, I fear that 3D Coat will put off a lot of experienced artists and when the voxel technology catches on, it could spell trouble for Pilgway. This software has a lot of potential but at some point that potential has to become consistently stable and predictable enough to be production ready if its to be taken seriously and used by professionals.
Posted 04 April 2012 - 08:21 AM
I could cry seing those pics. Not because they're spot on, but because these issues have been raised so many time with the resultat that we know I can't take it anymore and decided to move on.
Thank you John for trying to make the words cross the border, I hope you're sucessful, cause honestly I abandonned a while ago. I'll come back to see if this goes somewhere.
This software reminds me of silo, so much potential, but no long term view...
At least Silo still has a chance now that T.y.R.R got hands on code:
Whoa - where did you find this out about Silo?!
So I'm a long time 3d modeler, pro and lately not pro (I work fulltime in the game industry as a designer right now) - I've literally used every modeling product since 3dstudio r4 (except blender, yuck) and photoshop since version 2.
Just an observation about this thread - this improving the brushes stuff has been literally talked about since 2 years ago.
Granted, the optimizations have helped a lot. I can get more detail out of a lower resolution mesh and the brushes are a lot more responsive. But I can't get consistent results using ANY of the toolset. I just find myself flipping through the brush types, tweaking settings and hitting undo a lot trying to do things that are no-brainers in zbrush. Nevermind the bazillion usability issues, most of them small, but they add up and when I get inevitably frustrated I usually head back to zbrush and dynamesh and suck up the weird interface paradigm.
I would still never be able to finish a sculpt in 3dcoat, the voxel resolution demands and the brushes would literally drive me insane to attain the zbrush 'crispness' and LoD needed.
3dcoat needs serious battle-testing and a group of pro/semi-pro users dedicated to creating projects in 3dcoat who can help Andrew create a roadmap and drill down on what's essential to get 3dcoat to the next level of USABILITY. Listening to the squeaky wheels in the forums will only confuse the long-term direction and architecture of this product. Like what was the last major update? Painting on voxels? Gah!
As a long-time silo user, it's heartbreaking because the parallels between the two are staggering. Both products have/had so much potential to beat out the current industry 'giants' but both seem to have focused on short-term sales goals instead of a proper product architecture.
Bless Andrew and his herculean work ethic - he's trying to support a product that has a feature list a mile long. But he needs to focus and FINISH one area, get it usable by actual artists, then move on to the next area of the product.
There's still hope, I think. At least the finished work gallery submissions are showing potential.
0 user(s) are reading this topic
0 members, 0 guests, 0 anonymous users