3D-Coat 3.0 ALPHA

[Deadman21]

While compiling OSX version we have found an interesting bug that caused crashes on not very new video cards. So new version for Win will work defenitely more stable on PC-s that had problem with running 3DC.

About manipulator - idea is that it will be stretched between two points inspace. You will be able to move pivots or transpose using icons.

Speaking of crashing. The GL version still crashes on my PC dell XPS Nvidia 7950 when I go into the volumetric sculpting area. Is this happening on other systems.

[Andrew Shpagin]

Speaking of crashing. The GL version still crashes on my PC dell XPS Nvidia 7950 when I go into the volumetric sculpting area. Is this happening on other systems.

This point was fixed today (I hope) during preparing OSX version.

[ldzywsj]

Andrew,in v35 when I merge a obj file to VS mode and then to sculpt it ,it disappears,is it a bug?

[Deadman21]

One critisim or suggestion or whatever you wan to call it

I really liked the transpose methods you already have in the other section of 3d-coat. For continuity why would you want to totally recreate the posing method unless you plan on reworking that area as well. Whatever method you choose should be the same in both areas. Again personally I like the old method

[Deadman21]

Andrew,in v35 when I merge a obj file to VS mode and then to sculpt it ,it disappears,is it a bug?

I just tested this and did not have a problem with it disappearing. It comes in small and I need to scale it up and then of course press enter to merg it.

[ldzywsj]

One critisim or suggestion or whatever you wan to call it

I really liked the transpose methods you already have in the other section of 3d-coat. For continuity why would you want to totally recreate the posing method unless you plan on reworking that area as well. Whatever method you choose should be the same in both areas. Again personally I like the old method

Yes,yes,I have the same sense,in sclupting mode,there is "draging points",It is a transpose tool,I love the method too! but it also should be improved

[ldzywsj]

I just tested this and did not have a problem with it disappearing. It comes in small and I need to scale it up and then of course press enter to merg it.

Sorry,it is my fault,I didn't press enter.thank you Deadman21.

[3DArtist]

Personally i like the silo gizmo, clean and straight forward. I think the gizmo in 3DC is a little confusing to use. One main thing that i dont like is when the scale boxes are not in alignment with the move arrows. Also in silo i can press a key, move and rotate the gizmo, press the key again to set the new orientation. quick and easy.

Having the floating palette is nice but i would also like to be able to do those things directly on the model.

[Deadman21]

Personally i like the silo gizmo, clean and straight forward. I think the gizmo in 3DC is a little confusing to use. One main thing that i dont like is when the scale boxes are not in alignment with the move arrows. Also in silo i can press a key, move and rotate the gizmo, press the key again to set the new orientation. quick and easy.

Having the floating palette is nice but i would also like to be able to do those things directly on the model.

A key to rotate the gizmo would be perfect. I have never used silo didnt know it had that feature.

[3dioot]

I spend some time on thinking out the gizmo's. Here is a (pretty refined) "mock up".

There were a few things i wanted to make certain.

1) each transpose had to be very different from the other one.

2) create recognition with current "standards" as much as possible without getting too hung up on current standards at the same time

3) it had to read well as contours

4) each gizmo needed to be a solid volume that was easy to click upon

I think i managed to do ok. :P

1) each transpose has a very different shape and profile. Only rotation has round shapes. Only scale has cubes. Only move has arrows/spikes.

2) I had to make some concessions here. Especially with rotate which normally consists of rings. This would be hard to click on and implement correctly (radius based on length of transpose tool etc)so i decided to go for a sphere (easy to click on) and a squashed cylinder (more solid version of a ring). Scale was the easiest and i think its very recognizable. Move was the hardest. I chose for the multidirectional spiked form because it was very different then both rotate and scale and it was a solid form. Anything with multiple arrows got really messy and looked complicated, i did try it but it just didnt look good so i ended with the single volume. The single arrow for movement along the transpose reads very well and was an obvious choice.

3) i put a contour shot in the image as well as a view thats slightly turned after the transpose line has been placed. I think they read well enough as different shapes. Offcourse each transpose could get a different color to differentiate them more but im afraid this will get confusing with people thinking about axis constraints (xyz is usually depicted as rgb)

4) i kept to simple "full" shapes. Both for clarity and for the ease of clicking. Zbrush has -specifically- chosen not to use gizmos and use circles. The main reason is that these are fixed to screenspace and always easy to click upon. The sacrifice is that you cannot read them; transpose move, scale and rotate all look the same in zbrush. I think this sketch combines the best of both worlds. Easy reading and relatively easy to click upon. I assume that Andrew will make it so you can adjust gizmo size (like you can in pretty much every other 3d app that uses gizmos) so you can make the gizmo's a little bit bigger or smaller if you so desire.

The first and second point of the transpose lines are simple dots. No gimmicks there. Ofcourse move and selection should be one click. It frustrates me highly with the curve tool for example that i have to first click to select the curvepoint and then click again to move it. Its slow. If you click on the green dot it sticks to the cursor immediately, you can move it, on letting go of the mousebutton (or lifting the pen) the point gets placed in the new spot. Simple, fast and without fuss.

Im of for dinner now but will be back to discuss it further. I read all the posts and cant wait to talk some more about it. (i always need more time when im on this forum; its insane)

3dioot

PS

I hope its obvious what gizmo does what but just in case it wasnt obvious:

Rotate

Sphere: rotate in screenspace

Cylinder: twist along transpose line

Scale

Cube: uniform scaling

Long box: non uniform scaling along transpose line

Move:

Multidirectional box: move in screen space

Arrow: move along transpose line

[adrianr01]

a improvement of the current gizmo could be to have an on mouse-over colorstate, so you know a transform is active, it would help alot not clicking just next to it...

delete the black cirkel, it doesnt do anything but blocking the rotation circle and makes it heavier

[philnolan3d]

I think the gizmo in 3DC is a little confusing to use.

I will agree with that. It's got too many things sticking out of it

[tree321]

With transpose can you add select all option?That way we could draw out separate objects and move them or just move whole forms.

The other thing I was thinking of is drawing shapes with the line and curve tools in the brush pallet.That way we could mask out more complex shapes and transpose the selection.

[cyartist]

You guys can debate this all you want but this is the wrong way to do transpose on a advanced level.

If I was Andrew I would set this as the standard (see link below) not a ZBrush-like Transpose.

http://www.kunzhou.net/publications/MeshPuppetry.wmv

This video example shows the abiltiy to make a rig as well. Which is sorely needed in 3d Coat.

Andrew could solve two limitations of 3d Coat in one Feature.

[splodge]

You guys can debate this all you want but this is the wrong way to do transpose on a advanced level.

If I was Andrew I would set this as the standard (see link below) not a ZBrush-like Transpose.

http://www.kunzhou.net/publications/MeshPuppetry.wmv

This video example shows the abiltiy to make a rig as well. Which is sorely needed in 3d Coat.

Andrew could solve two limitations of 3d Coat in one Feature.

That's all nice and stuff, but sometimes we just want to bend something and not spend a few hours creating a skeletal structure with limits and balancing points.

I think the current Transpose is great for now and very easy to use. Maybe Andrew could make a skeletal based transpose later. But I think the current simple version should always remain for those that just want to bend something.

[tree321]

You guys can debate this all you want but this is the wrong way to do transpose on a advanced level.

If I was Andrew I would set this as the standard (see link below) not a ZBrush-like Transpose.

http://www.kunzhou.net/publications/MeshPuppetry.wmv

This video example shows the abiltiy to make a rig as well. Which is sorely needed in 3d Coat.

Andrew could solve two limitations of 3d Coat in one Feature.

Yes your right.If this system could be implemented in 3DCoat that would be great ,but it might take a while, if its not patented or if Andrew can come up with an alternative.These vids have been shown before and Andrew is aware of it.Thanks for sharing.

[cyartist]

That's all nice and stuff, but sometimes we just want to bend something and not spend a few hours creating a skeletal structure with limits and balancing points.

I think the current Transpose is great for now and very easy to use. Maybe Andrew could make a skeletal based transpose later.

But I think the current simple version should always remain for those that just want to bend something.

Actually you can build it out a small bone for small areas or a whole skelteton if you want a full rig.

Be more open minded about it........see the longer bigger picture was my point.

[Psmith]

Puppetry & Posing

Although trying to rush or push Andrew is the last thing I want to do, I can't help hoping that he will catch our enthusiasm about this method of "puppetry". For many of us, it is exactly what we have been waiting for, literally, a decade or more. If this functionality just happened to appear in 3D-Coat, it would make it absolutely the most unique, artist friendly, most fantastic, revolutionary, state of the art, mind blowing, knock your socks off, obliterate the competition application that has ever hit the airwaves.

The solution to every character animator's nightmares and the fulfillment of every character animator's dreams. Simplicity, elegance and fun. I mention animation because there would be no sense incorporating this kind of advanced and intuitive system, just for posing of static characters. This system begs to be used to make those static characters come to "life".

And fierce competition, duplication of functions and early adoption of revolutionary technology are becoming the name of the software game for all 3D developers. I do believe Andrew can make this happen, but he has got to get "the easy things" that everyone expects out of the way first.

Psmith

[Andrew Shpagin]

I spend some time on thinking out the gizmo's. Here is a (pretty refined) "mock up".

Good idea! What about combining all that stuff on one axe?

[bwtr]

Carrara gives four Gizmo choices.

This, the universal one, is as good as any I have used on any app.

Brian

[Andrew Shpagin]

Updated to V36.

I have add selection with alpha, but it is slightly unfinished (undo works not perfectly with selection). But anyway I have made this update because there was fixed 2 very important longstanding bugs.

- bug that causes slowdowns after some time, cracked quads, crashes without possibility to save result. At least it was fixed. Thank to wailingmonkey who has given me so full descriprion of the bug that I have fixed it immediately

- compatibility/rener speed bug fixed.

I will be away almost all day so I decided to upload it now mainly because of important bugfixes.

[tree321]

Updated to V36.

I have add selection with alpha, but it is slightly unfinished (undo works not perfectly with selection). But anyway I have made this update because there was fixed 2 very important longstanding bugs.

- bug that causes slowdowns after some time, cracked quads, crashes without possibility to save result. At least it was fixed. Thank to wailingmonkey who has given me so full descriprion of the bug that I have fixed it immediately

- compatibility/rener speed bug fixed.

I will be away almost all day so I decided to upload it now mainly because of important bugfixes.

Thanks again Andrew!Paint selection is what I was looking for.Drawing with splines or curves would be a nice addition for more precise control over shape of masked area but this will work.Have a good day.

[Mantis]

@3dioot:

Your Handles are really great they are pretty!

I like the way you did it,

Maybe the rotate handle need a way to rotate on its own axis and the 2 others axis.

I think it needs a way to move it around like you do with the center circle in Zbrush.

And a question, how does the line is being build?

Does it find the center of your volume and it goes there?

But what you did is really great and readable, good job.

[ldzywsj]

Andrew,the V36 really works more smooth than V35,it is great!

[3dioot]

@ Andrew

Im glad to hear you like my idea.

"What about combining all that stuff on one axe?"

Im a bit surprised you are asking me this since in the same post your refering to i mentioned my strong dislike for combining all transforms in one gizmo. Are you trying to create a dilemma for me?

Lets see what the cons are to combining all:

1) you lose the recognition of different shapes for the gizmos.

Since there is no different state belonging to each of the gizmos you will always see all of them at once. This makes remembering what specific purpose each one has harder.

2) it takes more room

Obvious one really. The point is you should strive for an interface that has minimal "junk" in the viewport except for the sculpt. I dont want to see rotate or scale, or have the choice to use them, when all im doing is moving.

3) this would force you to implement a random order to the transforms gizmo's.

Right now there is a deliberate, chosen order to the gizmo's. Gizmo on the end is transformation based on screenspace. Gizmo below is transform based on the transpose line. This is the same for all three transforms. You will throw this order (which has alot of value when it comes to ease of use) out the window if you implement all of them on the same transpose line.

4) You cannot hotkey checkboxes.

I cant hotkey move, rotate or scale if they are just toggable with a checkbox. Its just a bad workflow mechanic.

5) it will start to look like food on a stick!

Lets look at the pro's:

1)you dont have to choose from three tools. Instead you only have one which would lead to one hotkey instead of three. I can hear the people that think they can sculpt with a tablet without ever touching the keyboard cheer allready in favor of this argument.

While its nice to go for an approach where the reliability on a keyboard is minimal it can and should never be a goal on itself. At the end you want to create a program thats as fast and easy to use as possible. Everyone who uses a 3d program intensely (one could say commercially) has one hand over the keyboard. It simply means you are more efficient since you use both your hands. While this doesnt mean tools should be fragmented over a thousand hotkeys it does mean that if there is something to gain by splitting a tool you should not hesitate because it would add a hotkey or two.

-------

So in general what is there to gain by splitting transpose up over its three transforms?

- Clarity (gizmo's being unique and their order having meaning)

- Ability to hotkey

- Less clutter on screen (obvious one really)

If you are still hellbent on throwing all three transforms on one transpose line im willing to think about it (reluctantly) but im afraid you will get something alot more cryptic, more symbolic then what i have suggested so far. Really.. Its not desirable. :s

-----

Thanks for the update. Great to hear you finally got those nasty bugs. While paint with pen makes it infinitely more usable there is still alot to say for a unified masking approach. Ill save that for later though. Its not my hobby to write essays.

@Mantis

Thankyou Mantis.

"Maybe the rotate handle need a way to rotate on its own axis and the 2 others axis."

You get two axis with the current approach. You get a z-axis based on the depth in screenspace which is presented by the outer gizmo (the sphere) and you get an axis along the transpose line which is presented by the second gizmo (the flattened cylinder). The only axis you could argue you are not getting is the one perpendicular to the transpose line and parallel to xy screen space.

Look at the picture in the lower left. It shows clearly the two axis you have available to you. The only axis you could request is the one that would allow you to rate towards and away from you.

You could argue that one "axis" of the transform is missing. The problem is that you need to look at all of this from a somewhat different perspective.

First a picture from 3dsmax. What you see here (i dont know if your familiar with 3dsmax) is two flyouts. The first one shows all the different coordinate systems you can pick in max. You can see you have a wide choice. These do nothing else but dictate what direction the different axis point in. You could say the orientation of your transform gizmo before you have done anything with it. Next to it is the flyout for the pivot point. As you can see you get a wide range again. Not only that but you actually have more choice then whats shown here. It was not untill recently that 3dsmax implemented a "working pivot". Nothing but a fancy name for being able to freely place the origin of your transform. Nowadays transform gizmo's are all the rage but i remember when i still had to enable -each- specific axis i wanted to work on. From my head it was function keys F5 to F8 enabling, x, y, z and the F8 being a togle between the three different planes (xy,zy,zx). You may know all of this but stick with me a bit longer.

While i dont think we are copying zbrush's transpose (we did improve on it allready in fact) i am going to refer to zbrush now to illustrate a few things just because its easier to make a point with something thats allready implemented and "real" then compared to some hypothetic example.

The genius of "transpose" is that its essentially nothing new (really it isnt). It is the ultimate example of the saying that less is more. Instead of thinking from a perspective that you had to be able to transform in every way possible, with every coordinate system imaginable, and with all choices available to you they reversed the approach. What is it you really want to do? And how do you -usually- do it? Now that we know that; can we streamline that process for you?

Lets start with the origin. You know where you want it. So why not do away with all options except one; you click where you want the origin to be. Projected from screenspace; first sculpt it hits thats where its placed. Thats one flyout completely gone. No choices; one simple and fast action to get the origin where you want it to be.

Coordinate system. Since you are sculpting you really only want a user defined coordinate system (stuff will never line up to world axis). Normally you translate over one specific axis. At most you translate over two. One axis is a line. Hmmm that solvable by having the user draw a line from the origin which he allready defined to some other point. The direction can be one axis.

Thats a good idea but what if the user wants to move in a plane? Everyone who has modelled knows that when you want to judge space what do you do? You rotate your view so you get as perpendicular to the movement your are doing as possible. You dont look at something that you want to move along its line of its movement. You watch it from "the side" so you can properly judge the distance you will move it. This will also help you place it "on" or "inside" whatever else you are modelling it. This holds just as true for rotate as for scale. You dont squash something while looking at it from above. You look at it from the side so you can judge what your doing. But wait; if people usually watch it from the side we can use screenspace as the coordinate system for planar movement! Looks like all bases are covered

That is "transpose" in a nutshell. There is nothing to it that was not possible before. I could use screenspace in 3dsmax (and often did). I could define my own user coordinate system. All that "transpose" is, is alot of options that are left out and a really smart wrapper for what remained. Smart wrapper = gizmo.

-----

Why this long story? Because as i said you could argue that (to stick with rotation) you cant rotate towards or away from your point of view but that it doesnt really matter because its highly unlikely you, yourself, would ever want that. Simply because it will be painfull to judge what your doing. Your first reflex would be to swing the camera round to look at it from the side. What that would mean is that suddenly the sphere (which was for rotating in screenspace) now allows you to rotate along the axis you envisioned when you looked at it straight on. This is so intuitive that i think most people that use transpose in zbrush never even realise they "miss" an axis compared to their complicated fullblown gizmo's from max/maya/modo/whatever. You realise it because Andrew first used the oldschool gizmo for "transpose" in 3dcoat before thinking about what would be a more elegant solution.

"I think it needs a way to move it around like you do with the center circle in Zbrush."

The flattened cylinder gizmo under the ball gizmo does exactly that (twist).

"And a question, how does the line is being build?"

This is why ive been saying its such a bad idea to have transpose include selection.

Because if you would not use those two points that are in transpose for making a selection then..... you could use those points to set the transpose line. Since its called the transpose tool it would make alot more sense to have that functionality inside transpose and selection outside. The way it is now the workflow would be something like this:

- Enter transpose

- Make selection

- Realise transpose lines are all wrong (they were right for making a selection but not for defining origin or direction)

- Drag your beginning and endpoint so they are correct

- Do your transform

What makes this worse is that every action you do to tweak your intial selection influences your origin and direction. For example adding or removing will set a new origin and a new direction!

Do you see how horrible that is? :lol:

Now what you suggested earlier made me think. Its not a bad idea to "replace" masking with selecting so that what you paint gets influenced instead of what you paint gets frozen (ofcourse still with the option to invert the selection). If Andrew then implements "painting influence" (the line, sphere, ring and maybe others) within 3dcoats selection method there is no reason to keep selection inside transpose. Try to imagine how that would work and what problems it would solve. It would be very flexible and it would not have you perform "double actions" like you would have to do with the current implementation.

3dioot

[buqa]

Lets see what the cons are to combining all

In fact, I'm against "universal" gizmo and for unique ones.

Why? Thats simple.

In the beginning, you have to ask, how much information should be in perfect case sticked to Gizmo. And the answer is: "A lot."

You should have icons for Move, Rotate ond Scale, thats obvious, but you shoul have icons for planar Move, Scale and Rotate (XY, XZ, YZ) as well asi for uniform Scale a free Rotate and Move. Informatoin abou how much (degrees, units, percent) you affected your model would be great as well. And maybe something I can't remember just now.

Probably its not possible to place all these informations on one Gizmo, so you have to have more states (or get over the loss of opitons, but its no good) a you can divide the state of gizmo in few ways. Division between separated Move, Rotate and Scale is imho most logical.

But if somebody thinks out how to get all these things into one well arranged gizmo, I wouldn't be against.

And 3dioot is right, maybe it's not necessary to have all three axes when using only screen space, but depends on how much should 3DCoat be modeler and how much ... just Zbrush. Andrews choice.

[Andrew Shpagin]

Ok, I agree, but maybe we can add some hot point on the line to call menu and switch between move/rotate/scale faster?

[Andrew Shpagin]

At least I have installed Vista 64. It was not easy. It hanged many times during installation... At the and I have disconnected all devices and it installed. But I am still not very glad with it... wifi isvia PCI card is not working...

[philnolan3d]

At least I have installed Vista 64. It was not easy. It hanged many times during installation... At the and I have disconnected all devices and it installed. But I am still not very glad with it... wifi isvia PCI card is not working...

I'm surprised, I've had no problems with Vista 64 since I installed it just over a year ago.

[Mantis]

Was thinking about something about posing and topology.

If you want to pose your character in 3DCoat and sculpt it afterwards, you need a way to move the topology while posing your character.

It's easier to do a topology while in a relax pose, in a T-pose way.

Making a topology while your model is posed can lead to an hard time while you will retopologize it.

You will even not be able to use symmetry mode because of your pose.

I think that one workaround would be to store a Tpose of your sculpted model, like that when you finish your sculpt you come back to your Tpose, do the topology, than when you come back to your pose the topology fit to it.

It is a must have because you have no way to get a topology linked to your model since you are modeling in voxel and it can end in some problem if you want to render it withouth having to use a raw mesh.