Doom miniatures as 3d models?

[Blastfrog]

Wanted to post this in the Doom miniatures thread, but didn't want to derail it. So, you know how it's possible to scan real objects in as 3d models? What if one were to use the Doom miniatures as a base to creating much better models (the JDoom ones suck ass)?

I personally prefer sprites, for the record, but I can imagine this would be pretty cool to use in a modern engine. (maybe Doom 3/4?)

Anyway, I was just wondering what you guys thought of this idea, and if it's popular enough, perhaps set something up to scan the miniatures to make models from them?

[yellowmadness54]

well, weapons, death animations, other animations like firing, all would be much harder to do.

I dont see this as working.

[Blastfrog]

yellowmadness54 said:

well, weapons, death animations, other animations like firing, all would be much harder to do.

I dont see this as working.

What? Obviously all you'd need to do is repose it to be in a T-pose, and give it a bone structure to move the pieces around. Just throw in a weapon model (if needed), and you're good to go.

[yellowmadness54]

Sodaholic said:

What? Obviously all you'd need to do is repose it to be in a T-pose, and give it a bone structure to move the pieces around. Just throw in a weapon model (if needed), and you're good to go.

Ok, I guess that works. But still, the monsters will be moving in ways that that would be hard to do. And that still doesnt fix the death's of them or other movements.

It may work for Imps or zombies, but something complex like a spider mastermind that's brain melts and collapses upon its self would be near impossible.

[Xaser]

Hmm, scanning things as 3D models? You mean, as in create a 3D model out of a scanned figurine? Sounds hella hard to do, if that's what you mean. Unless you're that 3D Kinect guy. :P

[Blastfrog]

yellowmadness54 said:

Ok, I guess that works. But still, the monsters will be moving in ways that that would be hard to do. And that still doesnt fix the death's of them or other movements.

Um, actually it wouldn't be harder to make death and firing than just walking. I'm sorry, but I don't see the logic in your argument. You should study how 3d models work, because no offense, but you really don't seem to understand the concept well enough to make these kind of statements.

yellowmadness54 said:

It may work for Imps or zombies, but something complex like a spider mastermind that's brain melts and collapses upon its self would be near impossible.

It wouldn't be that hard to make the spider mastermind, honestly.

Xaser said:

Hmm, scanning things as 3D models? You mean, as in create a 3D model out of a scanned figurine? Sounds hella hard to do, if that's what you mean. Unless you're that 3D Kinect guy. :P

What I had in mind is one of those 3d barcode scanner softwares, where you just use a webcam, a barcode scanner (or something functionally equivilent, only need the red bar of light), and a 90 degree corner. Just scan the bar across the object, and it generates a 3d model.

After that, it needs a little bit of manual editing to make it usable, but that wouldn't be too difficult.

[Mithran Denizen]

The 3D scanning hardware/software that I've worked with in the past was overpriced and/or overcomplicated garbage that left much to be desired, and it was almost always just as much work to get an accurate model of something as it would've been to build it from scratch.

That wasn't dealing with anything intricately detailed like the Doom miniatures, though, and it was years ago, so feel free to prove my pessimism wrong. I haven't really followed how the technology has progressed at all in this decade.

It still would be a pain in the ass to rig and animate them well though, which is the worst problem with most of the available 3D monster models for Doom.

[Technician]

Fuck this, I want the models used for Doom 64.

[DaniJ]

Given access to a half-decent 3d model scanner and modern modelling software and this wouldn't be all that complicated. First put the scanned point field data through a tool which turns it into a poly mesh (take a nap, it'll be churning for a while...). Once achieved you'll have an incredibly dense mesh, so the next step is to use a re-topology tool to simplify it sufficiently for further editing (take a walk?). Now you have a clean mesh, import it into your 3d app of choice and pose it in a relaxed position for rigging, skinning and then ultimately animating.

The technology has improved dramatically in recent times and is now used by Hollywood for scanning real-life assets, like actors, for CGI work.

However I have to ask - why go to all this trouble for the Doom miniatures? ...seems to me that the only reason this was brought up is because the existing 3d models aren't liked.

Personally I think it would be much easier for an accomplished model artist to create new models from scratch, rather than going through all that. After all, its not like these need to be indistinguishable to real-life counterparts.

[printz]

What about using some laser-pointing device to scan the shape of the object?

[hex11]

Could be used as a means to get high-quality sprites for Freedoom or a TC. Find a cool-looking toy/model that can be animated (not a static mini) and hook it up to a rig like id used back in the day to take pictures of their clay models. Get all the shots for every walking/firing/death position, and then import into image editing software to touch up, etc.

[Maes]

hex11 said:

Get all the shots for every walking/firing/death position, and then import into image editing software to touch up, etc.

It's exactly that post-editing step that makes scanning kinda pointless, and not all that different from traditional stop-motion, multi-angle photography.

What Sodaholic wanted was a means to scan a model in full 3D, which is impossible to do with the quality required unless you have access to a dedicated 3D scanner such as the ones you might find in a major industry or an university research lab.

"Bubblegum & shoestring" methods such as using Kinect scanning, silhouette-based 3D reconstruction etc. will generate data of too low quality which will require just as much manual labor as making the models from scratch, to be usable.

Even assuming you have a full-quality 3D model of any given monster, using skeleton-based animation may work well for walking and attacking (assuming that the models themselves don't get significantly altered during those actions), but this is not true in all cases, e.g. during death many models actually break apart, split open and change frame structure and texture appearance dramatically unless you cleverly "fold" them (e.g. if you model the guts of a cacodemon too, and have them turn inside out during death).

And, let's not forget one final thing: let's say that you solve all of the previous problems. What source port will use your funderfull full 3D models? Will they be used directly as 3D models (in which case, bye-bye performance unless you keep to Doom 3-like numbers), or you will downgrade them to sprites anyway, with a limited number of rotations, Doom palette, and a limited -even if better than vanilla- resolution? Good luck getting them to work across different ports, too -in terms of what's available today, you'd have to use ZDoom and painstakingly create DECORATE overrides for the standard monsters, for any given resolution e.g. "2x resolution, 8-angle high quality monster pack", "3x resolution, 16-angle high quality monster pack" etc. etc.

And let's not even speak of coloring, palettes, and getting to use truecolor on flattened sprites....

[printz]

Maes said:

Will they be used directly as 3D models (in which case, bye-bye performance unless you keep to Doom 3-like numbers),

Can't the level of detail be scaled down?

[Maes]

printz said:

Can't the level of detail be scaled down?

Sodaholic said:

the JDoom ones suck ass

...and we're back to square one ;-)

Just how much should you scale them down? E.g. enough to keep vanilla maps playable, or to keep HR or NUTS.WAD playable? And once again, even if you have perfectly executed 3D models with Half-Life or Doom 3 quality skeleton animation, you will still have a solution in search of a problem (namely, a source port that can use them), otherwise you'll have to downgrade them to ZDoom DECORATE sprites, or even to vanilla resolution to actually use them.

[Vermil]

Zdoom and Decorate are irelevant in this thread; numerous ports support hi-res true colour sprites with infinite numbers of states in Wad, PK3, Zip etc

Certinally, both the MD2 and MD3 model formats are capable of higher poly models than the current "JDoom models" as people seem to be calling the current Doom models. Just nobody has made them and more detailed models will of course degrade performance.

[Maes]

Vermil said:

Zdoom and Decorate are irelevant in this thread; numerous ports support hi-res true colour sprites with infinite numbers of states in Wad, PK3, Zip etc

It can't be "irelevant" if it's part of the (few) ports that DO support such sprites, now, can it?

Don't confuse hi-res TEXTURES with hi-res SPRITES, also. In general, true-colour sprites are only usable in OpenGL or hardware accelerated ports due to their incompatibility with the vanilla lighting system (e.g. ZDoom and prBoom in software mode just don't support mixing truecolor and indexed color resources, only a more limited form e.g. truecolor skybox plus 8-bit textures / sprites. PrBoom+ does support some form of beautified/multiscaled textures but again, in its own formats (some custom textures in longdays.wad comes to mind).

Also, name me one hi-res sprite pack that is NOT bound to the specifics of one particular port to work.

[Vermil]

I'm not.

Doomsday and Risen3D for instance, support hi-res true colour replacements for any graphic, be they sprites, textures, whatever. Neither use Decorate and supported hi-res true colour replacements of any graphic before Decorate even existed.

[hex11]

So then, why did Adrian Carmack go through the trouble of sculpting clay figures, and have them photographed by a pro contractor? He could have just opened up his paint program and started drawing.

If it's so easy to make good-looking sprites from scratch, why are there so few (good-looking) monster replacements? Most of the decent ones are actually based on the original monsters from Doom, Doom II, Heretic, Hexen, or Strife.

These might be dumb questions, I don't know. I'm not much of an artist, and haven't done any pixel art since the mid 90's (used to play around with Deluxe Paint on the Amiga, good stuff...)

No argument about 3D models... If I wanted that, I'd go play Quake or something. ;)

[Maes]

Vermil said:

I'm not.

Doomsday and Risen3D for instance, support hi-res true colour replacements for any graphic, be they sprites, textures, whatever. Neither use Decorate.

This one clears up the prBoom+ situation:

http://www.doomworld.com/vb/source-ports/52944-how-do-i-get-high-resolution-sprites-in-prboom/

The software rendered version doesn't even support "plain" high resolution, let alone true color. It only works with hardware acceleration. The other ports you mentioned are hardware acceleration-only affairs, AFAIK, and I am not aware if they support the same HI_START/HI_END marker lumps as (gl)boom+, or they require their own particular extensions.

So yeah, OK, there are ports that support hi-res sprites, but that's not what I contended. What I said was that you can't support them all in an unified way. And notice how we got to this point (sprites) starting off a post about 3D models. If hi-detail, hi-performance models will even come to be used, no matter how they are produced, they will probably become part of a special branch of a currently developed port, like the Voxel ZDoom project.

All of these projects have some problems in common:

• A lot of manual work involved to create (initial) content for them, usually only by one man.
  
• Specialized, usually "one off" engine/port.
  
• Reliance on hardware acceleration (with the exception of voxels).
  

What I said remains true: even if you (or someone else) comes up with a magically crafted and perfect collection of 3D models for ALL of Doom's monsters and sprites (and BTW, I think there was an actually pretty cool project in WADs & projects), getting them to work across ports will be another major undertaking on its own.

[tempun]

Maes said:

PrBoom+ does support some form of beautified/multiscaled textures but again, in its own formats (some custom textures in longdays.wad comes to mind).

This is a lie. PrBoom+ (in GL mode) supports hi-res textures/flats/sprites/HUD graphics/font, in PNG format (and probably others). Low-res version is still required though, to define the scale and for software mode. It uses no definition lumps whatsoever.

Maes said:

Also, name me one hi-res sprite pack that is NOT bound to the specifics of one particular port to work.

See above. You just need to place graphics in the proper place. HI_START/HI_END lumps is going to be a cross-port method; originally for ZDooms, PrBoom+ supports it, and Doomsday will support it eventually.

[Maes]

I linked to that prboom+/glboom+ thread in the post before yours. So OK, glboom+ supports the HI_START/HI_END markers with a scale implied by the hi/low resolution ratio, and you mentioned also ZDoom (probably GZDoom, as no software-rendered port supports true color and hi-res sprites). So of all existing Doom ports, only the OpenGL versions of two of the [FLAME_SHIELDS_ON] most popular [/FLAME_SHIELDS_OFF] support hi-res AND truecolor sprites for sure. I think we deduced that 2-3 posts ago. Can we please move on?

P.S.: interestingly, there's no port supporting just hi-res sprites without the truecolor component (e.g. just with vanilla 8-bit color), so hi-res also implies hardware acceleration.

Also, the HI_START/HI_END method has the drawback of having to include redundant low-res data as a reference for the hi-res data anyway, e.g. you can't have hi-res sprites at an arbitrary scale alone, unless you include at least some placeholder dummies indicating the equivalent "vanilla scale". DECORATE does get around this by explicitly requiring that you specify the scale for each frame, but it's a ZDoom only method, and a painstaking one at that.

[scalliano]

Technician said:

Fuck this, I want the models used for Doom 64.

+1

Besides, you'd only need to do the missing ones.

[Ed]

Technician said:

Fuck this, I want the models used for Doom 64.

Yes.

I'd like to remake every single monster with modelling clay and redo the entire sprite library in Gumby, stop animation fashion.. but I'm lazy.