ZBrushCentral

Very very annoying bug with Zbrush!

There is a “feature” in Zbrush where the user is supposed to be able to delete part of the faces in a mesh.
The bug is when you do your detail at the higher subD level is smoothed out considerably. This is a huge problem.
Because of the memory limits of Windows the user can not add enough polys to the mesh to complete details
for a full creature. Whats needed is to be able to delete a portion of the mesh in order to continue subdivideing
other areas to add further refinements.

In addition the user needs floating point maps so that they may delete a portion of a character, extract a map then
have that maps color values align to the displacement map of the deleted portion.

This is an advertised feature and should work.
Anyone else find this a problem or is it just me? If the poly limit is 8 million we need to be able to add details on
parts of a creature at a time! :confused:

oh dear i wish I could break 1.5m polys.

I wish I could help but really I just have some questions for you so that I can understand for later (haha i mean someday when I have a computer from this century).

are you a)wanting to delete portions of the mesh ie half a face rather than the whole face. or delete portions of a mesh like the torso kind of thing? and I am wondering why if it’s whole parts why you can’t? (I can’t make it past 5 subd’s so maybe thats a problem after 5? don’t know)

my second question is are you really looking to delete or hide portions to subd?

arghh I have more but I feel I am probably gonna irritate ya more than ya are at the moment. Forgive me but I am trying to understand some more about the creation and how it will affect perhaps what I might want to do with a model later and would rather get it right earlier rather than later.

You may want to check your preferences for your current subD limit problem.

But to be clear. Imagine if im working on a
male figure. I want to be able to subdivide
the model and detail the entire figure using
8million polys ( if I like ). Then I need to be able to switch back down to the lowres
and extract a set of ( float )displacement maps.

Lets say the head design requires more polys for additional details. I should be able to
delete the body of the figure except the head. Since my total scene/tool poly count is much lower this would free me to add more
subdivision levels.

Then if Ive avoided detailing the border edges of the head ( where the head and body meet) I would be able to extract new float dsp maps. Since the maps are float I could be
confident that the details for the head and body would match.

ahhh ok…next question…forgive me then if this seems stupid or obvious…

are you looking then to save the mesh with mt, do the delete…create displacements with the various parts then restore and use the displacements that way…or

create the individuals and multimark em etc later. …oh doh…you want the individual displacements for another prog…may I ask which one you use that would allow you to assign multiple displacements for one model (not much experience in this century with max, maya etc but this seems like a good thing to know)

edit…thanks for answering my q’s, I know your a busy guy. And yeah…I set my preferences but I have maxed my dinosaur in memory so usually I max that out to the point of pulling out hair around subd5. Not a z problem…an aminuts problem!

I think I got the idea of what you want to do now.

oh…another…is it really necessary to need 8 million polys? to get the level of detail you’re looking for? since I obviously don’t do film work or anything…I am wondering why such high poly counts would be needed and all I can think of is for movie or tv kinda work…and if thats the case…I really really can’t wait to see what you are up to with this!!
:smiley:

I am JUST starting to detail and Im at 4million+ poly because the creature is very long. Imagine if this guy had to look good up close! Now of course you would have a series of painted hiFreq displacements on
this as well. But I cant paint multiple maps on one surface in Zbrush. That’s a
bummer because it has such cool painting tools.

If the delete button did not lose detail and If I had FLOAT MAP output, I could work
the model in sections. And then have everything put back together at render time.

Aminuts,
About multiple maps on a surface, I think
most renders support this. Maya default render, Mental Ray, Prman…

Hi,

I get the same smoothing when deleting from a multi subd level model. I tried all the switches but to no avail. Morph targets do not help either as they are based on the original point count.

I found a work around but I am not sure that it will work for all situations. It is easier to let a script do the talking, especially since I remembered to “End Record” the bugger while testing. :wink:

Deleting without smoothing.TXT

The only real limitations are no use of triangles in the base mesh and the rules regarding visibility and subd levels. Edit: You can get away with triangles by deleting subd level 1 and hidding polygons from subd level 2(now subd lvl 1). But I dont know if that will cause you problems.

Nice looking creature.

Cool model you got going there, Skycastle. I hope that Tveyes work around works for ya cuz you sure are right about the smoothing, yikes!

Thanks for answering my questions.

HHAHHAHHHaaa !

I like it! That does indeed work for me!

That solves part of my problems. We still need floating point map output so that displacements will align between a mesh split into multiple parts.

This would expand Zbrush’s potential a lot
more! Instead of an 8million poly mesh you
could have 4 tools each with 8million polys
for one character!

I need this! :stuck_out_tongue:
Pixolator where are you?..

Thank you TVeyes

TVeyes - that’s brilliant !! Awesome workaround ! Skycastle - exceptional work on that model, beautiful detailing.

Thanks guys for the inspiration and fix !!!

Now if only there was an equally clever way of dealing with edgeloop distortions I’d be a really happy camper !

Skycastle - after reading the last post I did some thinking on the floating point map thing you discussed and wanted to understand a little better. I’ve noticed that when I have deleted some geometry and gone to a higher rez for addittional detail (without recreating UV’s)that when I create a new displacement map It gives me the full “old UV” map layout - now mostly grey - with the remaining high rez geometry areas showing up nicely.

I thought at the time that this might be usefull, you could use the same starting model, delete unwanted geometry, detail the remaining section, generate a disp. map then move on to the next area starting with the base model again. Eventually you’d have to composite the maps but they SHOULD all align properly if based on the same “starting” full set of UV’s.

Unfortunately I’ve deleted the test map in question and can’t confirm this right now - have you tried detailing out a section of the model and generating a displacement map with or without the same results using the full model set of UV’s ? If you stayed away from piece borders as stated it should work - is this what you were talking about ? Am I at all on track ?

Well

Right now Zbrush is calculating the displacement values in float but then remapping those values to fit in 16bit 0-1 color space.

That means the amount of remapping/scaling
that can occur in the map is diffrent depending on what details are being calculated.

With float map those values will be in ture
world space units. This would allow you to
calculate separate parts of a mesh independent from one another. In the end you
would be sure that the maps would align.

Then using Shake, Cinepaint or whatever the
user could take the maps and rescale(the values ) them to fit their needs.

But yes you are on the right track.

Got it.

Here’s a wacky thought - would it be worth trying to add a common detail “peg” to each section of the model that has a constant height (and depth now that I think about it) ? so there’s a way of “setting” the min/ max values for each section and force the remapping to be consistent ? If small enough, the “peg” areas could be painted out later…probably way more trouble than it’s worth but thought I’d throw it out there :slight_smile:

Skycastle,

Forgot to mention this last post, but is there any chance of seeing a wireframe of the base geometry for this creature (or at least the part shown above) ? I’d love to see how far the maya reworked model is taken before starting the displaced detail with your workflow.

<BLOCKQUOTE>quote:</font><HR>Here’s a wacky thought - would it be worth trying to add a common detail “peg” to each section of the model that has a constant height (and depth now that I think about it) ? so there’s a way of “setting” the min/ max values for each section and force the remapping to be consistent ? If small enough, the “peg” areas could be painted out later…probably way more trouble than it’s worth but thought I’d throw it out there<HR></BLOCKQUOTE>

If you did go this way, would it be easier to add a secondary part to the model that is entirely detached from the main part and has a discreet space in the UV map? It would then act as an isolated calibrated reference. You would obviously want to make it with as few polys as possible so as not to eat too much into your total poly count.

Also, if the maps are generated in a relative 0-1 colour space, would this mean that the reference object would have to have displacements that were equal to or larger than the main part of the model (if so, you wouldn’t want to make it too much larger otherwise you’ll be limiting the available colour space for the wanted part of the map)?

Yes you have the right idea on the UV tiles.

As far as color space thats the whole idea for using Float maps, you have a huge range
in both positive and negative directions.

In production float displacement maps are used. But forget production even at home
I dont want to be limited by poly counts.

Thanks skycastle, I finally understand how having float maps would improve things other than potentially more detail than a 16 bit image.

Is there not a way to use ramps in your shading tree to control the strength or amplitude so you could make multiple displacement maps work together? or is that gonna soften and or clip details.

Also I always find your work very inspirational. I was wondering where, and what u use for reference when detailing your creatures. And is any of your work on the dvds gnomon is putting out with Meats and Bay?

Thanks Ken

SkyCastle. First off nice model.

One thing I’m doing that you might be able to do as well is spliting my model into multiple pieces inside 3dstudio. I’m assuming u plan on using automatic mapping. If so Just detach diffrent portions of your mesh and asign each a unique mat id. Then Uv each peice seperatly and export each as an .obj. Then in zbrush add your detail. After that export the displacment maps and when back in 3d studio or maya or whatever just restitch all the vertices. The only disadvantage is you’ll have to have a bunch of seperate image files for your model and your’ll have to worry about texture seams. If you split it in the right spots they should be easy enough to hide however. You may be to far along to do this now but you may want to try it in the future. Hope this helps :slight_smile:

Thanks.

Thats exactly what I would LIKE to do. The problem is because you cant export
float maps each map for each seperate part will have a different scale.
The maps just wont match up at the seams.
If you had float maps the user would be able to scale them to what ever bit depth they need( not everyone can render float data ). Zbrush calculates in float why not just give the user that data as an option? It the only way the
user can work the mesh in parts. Pixologic…

Ah. That does suck. I’m luckyenough to have a model where seams wont show as they’ll be hiddon from site but I see what you mean. too bad :frowning:

Bump !

I really think ths point would need further development. Is there any improvements, methods ? News from Pixologic ?