PCSX2 in stereoscopic 3D
#31
(04-17-2013, 07:59 PM)Fanoffemdom14 Wrote: Hah, and i remember some guy telling me that this was impossible.

Things don't need to be impossible to be bad, or less than ideal.

Real 3D is not just doubling and post processing frames, it involves displaced cameras taking the shots.

Talking only for myself, that's not 3D, that's a simulacrum, it is not something I buy, but again I'm speaking only for myself.

I was not joking about the headaches. 3D is not meant to bring a mess to the brain, actually it is still not good enough to avoid the mess... that even when is done with that purpose, with the best the nowadays technology has to offer.

Still, I believe it not being worse than that MS idea of lightpainting the wall... there are those who will buy it (both).
Imagination is where we are truly real
Reply

Sponsored links

#32
(04-17-2013, 09:08 PM)nosisab Ken Keleh Wrote: Real 3D is not just doubling and post processing frames, it involves displaced cameras taking the shots.

funny you mention that. could you post a screenshot of a side by side? i'm so lazy right now. Laugh

i didn't even compile it. i found it more interesting wasting time reading just the patch code. and thought "f*ck it, this' not gonna be the real deal". no math in it. i dunno much about that code but the essence of reprojection is commented out. there it goes. BS
Reply
#33
(04-17-2013, 11:34 PM)xstyla Wrote: funny you mention that. could you post a screenshot of a side by side? i'm so lazy right now. Laugh

i didn't even compile it. i found it more interesting wasting time reading just the patch code. and thought "f*ck it, this' not gonna be the real deal". no math in it. i dunno much about that code but the essence of reprojection is commented out. there it goes. BS

No need to go far, the thread title itself states stereoscopic, I guess what it means?

Is possible to achieve the illusion of 3D at images from "mono" source, like is possible to achieve spacial sensation with sounds, just is not the real thing. Anyway, if done right the whole thing could be interesting, done wrongly the result is a poor shadow of the former self.

Some things are best leaving as they are, least we do them more harm than good.

PS: Like the attempts on making Mona Lisa better by painting mustaches on her.
Imagination is where we are truly real
Reply
#34
(04-17-2013, 07:59 PM)Fanoffemdom14 Wrote: Hah, and i remember some guy telling me that this was impossible.

Because this isnt proper stereoscopic 3D. This is only anaglyph 3D. Proper stereoscopic 3D (which is what 3D tvs do) isnt going to happen with pcsx2.
[Image: jmt41l-2.png]
SSD: 250GB | HDD: 2TB | GPU: MSI GTX 970 4G Gaming
Reply
#35
(04-18-2013, 02:25 AM)Dante3hs Wrote: Because this isnt proper stereoscopic 3D. This is only anaglyph 3D. Proper stereoscopic 3D (which is what 3D tvs do) isnt going to happen with pcsx2.

you could do that too. it's just a cheap geometry shader to double the primitives to render 2 views and do the reprojection with 2 shifted eye matrices from an identity projection. clip and done. dunno if depthless sprites may be a ***** there tho. also you won't look around corners like that, cause the geometry is culled on the VU but it'd render different angles for every forward visible polygon, which is still stereoscopic.

i'll leave it to anybody else to try to code. i can't. i haven't enough games to check if it's a all around solution. so...
Reply
#36
(04-18-2013, 02:25 AM)Dante3hs Wrote: Because this isnt proper stereoscopic 3D. This is only anaglyph 3D. Proper stereoscopic 3D (which is what 3D tvs do) isnt going to happen with pcsx2.

Yeah

Sometimes I think some people become too enthusiastic with new technologies and want to convert everything to them irresponsibly.

Let's take the case of upscale, it is good thing, it makes the image better and easier to the eyes and too the brain. The same for shaders which can be good or plain bad taste but that's just it.

Now, this 3D thing is another story, it is potentially harmful if incorrectly done. The brain receives dissonant information and the result may be tiresome to say the least.

The real 3D content is designed, directed and edited to produce acceptable results, that is done take a take so to grant that quality. What the thread proposes is applying a method at the emulator level, whatever the content is being presented, the kind of thing you can show to your friend and get "WOW, that's great!" but at same time the thing nobody in common sense would endure for long.

Maybe someday, when the technology advances to the point the methods are refined enough to produce true 3D from scratch without becoming a pain in the long run... this day I may admit that something reasonable might come out from the attempts on recreating 3D from plain content.
Imagination is where we are truly real
Reply
#37
Real 3D content? The quality of stereoscopic 3D depends only on:

- 2 correct views, 3d objects, shaders and 3d world (without bugs and errors like shadows only in left eye, etc..)
- good seperation/convergence settings (adjustable/depends on user/eye seperation)
- output/display quality (ghosting)

As long as there are 3-dimensional objects and 2 virtual cameras/viewpoints, you get perfect stereoscopic 3D (2 images for each eye). As long as the 3D models are just that, Polygons, vertices, etc, it should work perfect in 3D. Of course, there can be shader problems, etc. But the basics are clear: 3D space, 3d objects (or even 2D "cutout" objects in a 3D space, think "Viewtiful Joe"), 2 views and you get stereoscopic 3D. As far as I understand, the PS2 does have 3D models, I think even the PSOne had 3D Games... Rolleyes.

How the 3d space is proyected in the renderer and emulation stuff, thats another question and may be a problem... I know nothing about programming and less of emulating, but as the PS2 does render 3d obejcts in a 3d space, it could be possible to render 2 views to get a stereoscopic output.

This has nothing to do with upscaling or getting 3d information from mono-sources, as in these cases, there is really data missing. The games on the PS2 on the other hand contain real 3d objects, modelled in 3dsmax oder something, and could produce perfect 3D. Now, as I said, maybe the emulation isnt capable of that because of projection mode or what-do-i-know... but to say, this approach is the same as upscaling or getting 3d from mono-pictures is just wrong, in my opinion.

I hope that one day this works with 3D Vision.... Cool Keep up the good work!
Reply
#38
(04-18-2013, 03:52 PM)soundstorm Wrote: Real 3D content? The quality of stereoscopic 3D depends only on:

- 2 correct views, 3d objects, shaders and 3d world (without bugs and errors like shadows only in left eye, etc..)
- good seperation/convergence settings (adjustable/depends on user/eye seperation)
- output/display quality (ghosting)

As long as there are 3-dimensional objects and 2 virtual cameras/viewpoints, you get perfect stereoscopic 3D (2 images for each eye). As long as the 3D models are just that, Polygons, vertices, etc, it should work perfect in 3D. Of course, there can be shader problems, etc. But the basics are clear: 3D space, 3d objects (or even 2D "cutout" objects in a 3D space, think "Viewtiful Joe"), 2 views and you get stereoscopic 3D. As far as I understand, the PS2 does have 3D models, I think even the PSOne had 3D Games... Rolleyes.

How the 3d space is proyected in the renderer and emulation stuff, thats another question and may be a problem... I know nothing about programming and less of emulating, but as the PS2 does render 3d obejcts in a 3d space, it could be possible to render 2 views to get a stereoscopic output.

This has nothing to do with upscaling or getting 3d information from mono-sources, as in these cases, there is really data missing. The games on the PS2 on the other hand contain real 3d objects, modelled in 3dsmax oder something, and could produce perfect 3D. Now, as I said, maybe the emulation isnt capable of that because of projection mode or what-do-i-know... but to say, this approach is the same as upscaling or getting 3d from mono-pictures is just wrong, in my opinion.

I hope that one day this works with 3D Vision.... Cool Keep up the good work!

That's right, yourself said it, but where is that correct 2 views, the 2 cameras view points in a PS2 game?

Well, all the rest was already spoken.

PS: Nature endowed us with two eyes and two ears for a reason. it's not enough one eye and you moving around. That two eyes perspective is what is tried to reproduce on actual 3D media content. Besides is what stereoscope means, otherwise is mono source.
Imagination is where we are truly real
Reply
#39
(04-18-2013, 03:56 PM)nosisab Ken Keleh Wrote: but where is that correct 2 views, the 2 cameras view points in a PS2 game?
Thats the part, that must be programmed. Smile No game on the market has 2 shifted cameras/views (except the games which come with 3D support programmed into it, which are very few, AVATAR is one).

But it's not a problem for all the DirectX Games out there (since DX8 and maybe older, i'm not sure right now), and it is not a problem for WII Games on Dolphin neither. It isn't even a problem for N64 Emulators or Google Earth...
Conclusion: As long as the game/app is in 3D, the "real 3d content" is there and if the emulation does allow it, a second view can/could be programmed. The game does not need to have 2 views natively.

I don't understand if this new patch really tries to create 2 views or not... But the game does certainly not need to have 2 cams to work in stereoscopic 3D. Thats the part of the driver (3D Vision, Tridef, IZ3D), patch or emulator.
Reply
#40
You create the views in the same way iz3D and the like do, messing with the rendered scene as soundstorm has said (you saved me lots of keystrokes Tongue2 ). The patch certainly looks like it's doing the work, and my quick look at the patch seems to confirm it. It's a LOT of work, and nice work at that.

Try it, then complain, otherwise go make another thread about how S3D is a gimmick, about how anaglyph ain't S3D or about what makes S3D propa' S3D.

#EDIT: While we're at it, Tridef does have a method for creating S3D from a mono source using the depth buffer. Wouldn't have the same quality, but I guess it might be lighter on resources. The method however, is listed as a way to avoid problems with post processings and shadows (like in the Unreal engine).
[Image: nbKSK.jpg]
Reply




Users browsing this thread: 1 Guest(s)