Quantcast

Page 173 of 183 FirstFirst ... 73123163169170171172173174175176177 ... LastLast
Results 2,581 to 2,595 of 2731

Thread: PS2 vs Dreamcast Graphics

  1. #2581
    Raging in the Streets Yharnamresident's Avatar
    Join Date
    May 2013
    Location
    British Columbia
    Posts
    3,769
    Rep Power
    62

    Default

    So whats the situation with ZeroGS?




    With the PS2 being so hard to emulate, there are some games that the ZeroGS' polycounter isn't compatible with. When this happens, it gives grossly low polycounts. What you do is press "F7" to show the wireframe, if the wireframe looks fine then you are getting accurate results. If the wireframe is a glitchy mess, then you aren't getting accurate results.


    If you look at the wireframe in Triggerheart Excelica, its a glitchy mess and runs sideways for some reason(and gives grossly low polycounts):



    Compared to Gradius V, the wireframe is fine and thus the polycounts seem right:





    You see this in other games. Test Drive Le Mans is a glitchy mess(with grossly low polycounts):



    Resident Evil: Code Veronica, same situation, glitchy mess and grossly low polycounts:




    Tokyo Xtreme Racer Zero. Wireframe is fine and thus the polycounts are right:



    MotoGP. This game is doing a ton more polygons than 22,000, no surprise its a glitchy mess:




    Ridge Racer V. Wireframe is fine and thus polycounts are what you'd expect:



    Suikoden III. Same situation:





    When documenting polycounts, you often gotta rely upon your own common sense to see if you are getting accurate results. Like look at this PS1 game, this is what 2300 polygons looks like:




    Clearly theres nothing on the PS2 that looks like that, despite ZeroGS giving those numbers for some games.


    I'll show one last example. Heres Radirgy saying 2200 polygons:



    That means that screen has less polygons than this screen in Gradius IV:



    But thats clearly not the case. Radirgy looks like a next-gen game compared to Gradius IV. Heres Radirgy polycounts on Dreamcast:





    So this is my unnecessarily large guide to the ZeroGS polycounter.
    Certified F-Zero GX fanboy

  2. #2582
    Hedgehog-in-Training Hedgehog-in-TrainingWildside Expert
    Join Date
    Aug 2018
    Posts
    243
    Rep Power
    6

    Default

    Quote Originally Posted by Yharnamresident View Post
    So whats the situation with ZeroGS?




    With the PS2 being so hard to emulate, there are some games that the ZeroGS' polycounter isn't compatible with. When this happens, it gives grossly low polycounts. What you do is press "F7" to show the wireframe, if the wireframe looks fine then you are getting accurate results. If the wireframe is a glitchy mess, then you aren't getting accurate results.


    If you look at the wireframe in Triggerheart Excelica, its a glitchy mess and runs sideways for some reason(and gives grossly low polycounts):

    [

    Compared to Gradius V, the wireframe is fine and thus the polycounts seem right:



    You see this in other games. Test Drive Le Mans is a glitchy mess(with grossly low polycounts):


    Resident Evil: Code Veronica, same situation, glitchy mess and grossly low polycounts:



    Tokyo Xtreme Racer Zero. Wireframe is fine and thus the polycounts are right:



    MotoGP. This game is doing a ton more polygons than 22,000, no surprise its a glitchy mess:



    Ridge Racer V. Wireframe is fine and thus polycounts are what you'd expect:



    Suikoden III. Same situation:



    When documenting polycounts, you often gotta rely upon your own common sense to see if you are getting accurate results. Like look at this PS1 game, this is what 2300 polygons looks like:



    Clearly theres nothing on the PS2 that looks like that, despite ZeroGS giving those numbers for some games.


    I'll show one last example. Heres Radirgy saying 2200 polygons:



    That means that screen has less polygons than this screen in Gradius IV:



    But thats clearly not the case. Radirgy looks like a next-gen game compared to Gradius IV. Heres Radirgy polycounts on Dreamcast:




    So this is my unnecessarily large guide to the ZeroGS polycounter.
    Your zerogs theory is disproven by my shots of trigger heart ps2. Because its not only zero gs but the obj dump as well giving those polygon count. You cant compare radirgy dc to ps2 .Your theory is wrong. If it has the same behavior as trigger heart ps2 that means they cull all the backfaces and keep polygons out of the frame to minimum via view frustum clipping(once i saw the obj dump i knew exactly what it was). If you look at my obj dump(trigger heart ps2) picture you will see it clips to very edge of the screen. The dc shows no such behavior and keeps going way past the "camera" .If you dont understand the clipping methods iam talking about please look them up, this is not a putdown as the ps2 uses them heavily as devs use the co processors for that.Basically the devs to checks against the camera to see which triangles are no facing the camera and dont render them , then they check via multiple methods what objects are not intersecting the camera field of view and then arent rendered. On the dc this waste cpu time so they just let the gpu auto cull BUT the cpu still has to submit to the gpu all those unseen triangles because they didnt clip it manually which is why we can see them.

    Zero gs polygons are correct.

    Triggerheart on the dreamcast, the terrain extends alot in all directions and is never clipped!


    Same scene on the ps2, the ps2 clips anything out of the view frustum and to take it further it backface culls the ships. If you dont trust me just do an obj dump and look, it will become clear:

  3. #2583
    The Future Is Yesterday Hedgehog-in-TrainingRaging in the Streets SegataS's Avatar
    Join Date
    Feb 2018
    Location
    My Chair
    Posts
    4,153
    Rep Power
    50

    Default

    Sorry to do this. I don't dislike PS2. I was just not a fan of how it rendered a lot of games. I feel the same way about 3DS. Something about the way 3DS game generally look is hideous to me.Also some of it is just to rag on AMD since PS2 can do no wrong in his eyes.



    Life!? ... What console is that on?

  4. #2584
    Raging in the Streets Yharnamresident's Avatar
    Join Date
    May 2013
    Location
    British Columbia
    Posts
    3,769
    Rep Power
    62

    Default

    Quote Originally Posted by Cloofoofoo View Post
    Your zerogs theory is disproven by my shots of trigger heart ps2. Because its not only zero gs but the obj dump as well giving those polygon count. You cant compare radirgy dc to ps2 .Your theory is wrong. If it has the same behavior as trigger heart ps2 that means they cull all the backfaces and keep polygons out of the frame to minimum via view frustum clipping(once i saw the obj dump i knew exactly what it was). If you look at my obj dump(trigger heart ps2) picture you will see it clips to very edge of the screen. The dc shows no such behavior and keeps going way past the "camera" .If you dont understand the clipping methods iam talking about please look them up, this is not a putdown as the ps2 uses them heavily as devs use the co processors for that.Basically the devs to checks against the camera to see which triangles are no facing the camera and dont render them , then they check via multiple methods what objects are not intersecting the camera field of view and then arent rendered. On the dc this waste cpu time so they just let the gpu auto cull BUT the cpu still has to submit to the gpu all those unseen triangles because they didnt clip it manually which is why we can see them.
    But you don't notice a pattern? How come every game that gives drastically low polycounts is a glitchly mess, and ones that give normal polycounts are fine? You ignored the elephant in the room.

    This glitch is also likely effecting the obj capture. Theres no way in hell those games are running at those polycounts, besides Gradius IV(which shows a proper wireframe).
    Certified F-Zero GX fanboy

  5. #2585
    Hedgehog-in-Training Hedgehog-in-TrainingWildside Expert
    Join Date
    Aug 2018
    Posts
    243
    Rep Power
    6

    Default

    Quote Originally Posted by Yharnamresident View Post
    But you don't notice a pattern? How come every game that gives drastically low polycounts is a glitchly mess, and ones that give normal polycounts are fine? You ignored the elephant in the room.

    This glitch is also likely effecting the obj capture. Theres no way in hell those games are running at those polycounts, besides Gradius IV(which shows a proper wireframe).
    The wireframe rendering is messed up at it renders multiple last frames on top of each other. There's no elephant in the room. You've never broken a game and ended up in a void where garbage texture is the last frame over and over? It's very similar to that but I have no idea why.

    The only glitch in the obj is the doubling which thanks to you I am no longer affected by. You honestly believe backface culling and view frustum clipping are glitches? No my friend, it's decision made by devs to save on fillrate and needless geometry calculations respectively. There a handful of Dreamcast games that do this but it's rare which is why you're probably confused. I knew since before you posted you were confused by these things I was just hoping you werent because it's the frame rate thing all over again.

    It's very cut and dry if you research the techniques and extract yourself you will see on the ps2 these technique is common. Not all games do it but alot do. It leaves very little room for doubt. Heck some games do one but not the other, it's all based on the optimization devs thought they needed.

    Extract, read up if you don't already know what they are and you will understand.

    Edit: if you still insist then I'll do radirgy one of these days and show you.

  6. #2586
    Raging in the Streets Yharnamresident's Avatar
    Join Date
    May 2013
    Location
    British Columbia
    Posts
    3,769
    Rep Power
    62

    Default

    Quote Originally Posted by Cloofoofoo View Post
    The wireframe rendering is messed up at it renders multiple last frames on top of each other. There's no elephant in the room. You've never broken a game and ended up in a void where garbage texture is the last frame over and over? It's very similar to that but I have no idea why.

    The only glitch in the obj is the doubling which thanks to you I am no longer affected by. You honestly believe backface culling and view frustum clipping are glitches? No my friend, it's decision made by devs to save on fillrate and needless geometry calculations respectively. There a handful of Dreamcast games that do this but it's rare which is why you're probably confused. I knew since before you posted you were confused by these things I was just hoping you werent because it's the frame rate thing all over again.

    It's very cut and dry if you research the techniques and extract yourself you will see on the ps2 these technique is common. Not all games do it but alot do. It leaves very little room for doubt. Heck some games do one but not the other, it's all based on the optimization devs thought they needed.

    Extract, read up if you don't already know what they are and you will understand.

    Edit: if you still insist then I'll do radirgy one of these days and show you.
    So are you saying with clipping techniques, they could get this Gradius V scene down to 4000 polygons? Thats what they did with the other games after all:

    Certified F-Zero GX fanboy

  7. #2587
    Hedgehog-in-Training Hedgehog-in-TrainingWildside Expert
    Join Date
    Aug 2018
    Posts
    243
    Rep Power
    6

    Default

    Quote Originally Posted by Yharnamresident View Post
    So are you saying with clipping techniques, they could get this Gradius V scene down to 4000 polygons? Thats what they did with the other games after all:

    Perhaps it can, I would need to see whats being rendered off and on screen. BUT the reality the devs did not use it because they had no need, simple. You may like trigger heart but I looked it over, nearly or more of its polygon count tends to be off screen and guess what? that works beautifully for the dreamcast architecture, To be honest triggerheart may be the only game that leaves that many polies unclipped offscreen on the dc. On the ps2 for what ever they said hey rendering all that terrain that isnt viewable by the players is waste and turned on a view frustum algo. Then they decided to take it further and get rid of the ships underside because you dont see them either way using backface cull. Its not the only one, resident evil 4 clips close to the viewport but does not backface cull the people (might call the stage)for example. I guess the count seems crazy but these methods can get rid of ALOT of geometry especially if alot is already out of the view port.

    Heres a short excerpt from the game swat from dev who worked on it. "The Xbox clipped geometry in hardware whereas the PS2 had to use software clipping, and would crash horribly if got it wrong and let it draw too far off the sides of the screen." Thats exactly what iam talking about. its nothing mystical or glitch or secret.I dont know how agressive swat is at clipping but it shows my point.
    Last edited by Cloofoofoo; 12-07-2019 at 02:34 AM.

  8. #2588
    ESWAT Veteran Team Andromeda's Avatar
    Join Date
    Jul 2010
    Location
    Wales, UK
    Posts
    5,298
    Rep Power
    58

    Default

    Quote Originally Posted by SegaAMD View Post

    frankly Dreamcast can't challenge ps2 in textures, who ever played The Matrix: path of neo, Tekken Tag, onimusha warlords, GT3, Jak 2, MGS2 knows what I'm talking about.
    MGR2? That is a hardly a game to show off PS2 textures and if listen to you the DC can't do grass. I'm sort of amazed you didn't bring up Silent Hill 3. That really is a game with great textures and also polygon beating performance to a DC game. But most of the time when it came to textures and clean high res display the DC had the PS2 beat IMO But if people play 'both' the PS2 and DC versions of DOA II, Grandia II, Rayman II, Code Veronica, Maken X.

    I really don't like the game, but since you talk of FPS's it's hard to beat the textures found in Outrigger and those in Daytona USA 2001. When the PS2 for me had the adv was in system memory and polygons. In that gen I always thought the XBox was the best.
    Panzer Dragoon Zwei is
    one of the best 3D shooting games available
    Presented for your pleasure

  9. #2589
    Hedgehog-in-Training Hedgehog-in-TrainingWildside Expert
    Join Date
    Aug 2018
    Posts
    243
    Rep Power
    6

    Default

    ok, Next is a comparison between Soul Calibur DC vs tekken 5 psp vs Soul Calibur Broken destiny psp. Surprisingly despite the age difference between both psp games and soul calibur dc their performance are actually pretty similar with the dc seemingly topping both. Both psp games LOOK more advanced because the art style and smart use of polygons on models make it Look nice(they focused on modeling the face with alot of detail but the overall body suffered.)

    Soul calibur DC cervantes vs seigfried : 28,611 tris per frame x 60 fps =1.7 mpps



    Soul Calibur dc : hwang vs yoshimitsu : 23,415 tris per frame x 60 fps = 1.4mpps




    Soul calibur broken destiny psp seigfried vs seungmina : 15,679 tris per frame x 60 fps = 940k pps


    Cervantes vs yoshimitsu = 16,845 tris per frame x 60 fps = 1mpps



    Tekken5 PSP lei vs xaioyu = 17,231 x 60 fps = 1mpps


    Hwuang vs asuka( someone mentioned the stage with the swaying grass? turns out the windswept effect is a very simple transparent layer with an animated "green wind" texture, very extremely simple similar to the dark vortex in skies of arcadia.) 8,756 tris per frame x 60 fps = 525K pps.

  10. #2590
    Hedgehog-in-Training Hedgehog-in-TrainingWildside Expert
    Join Date
    Aug 2018
    Posts
    243
    Rep Power
    6

    Default

    Another comparison , Radirgy ps2 vs dreamcast. Both games sport the same assets BUT rendering behavior is extremely different. On the ps2 the game does not backface cull the ships but it does view frustum clipping. Meaning unseen meshes will be tested for visibility against the camera and discarded if tested not visible.On the dc they did no such thing because of how the gpu works and how simply it doesnt draw beyond the viewports and most likely draws less than what the cpu sends. BUT the cpu is still calculating those models outside of view because they didnt manually clip before sending to the gpu, so technically alot shmups have inflated polygons count on the dc because whats not seen is being shows to us once we extract. I suspect the reason for this its because its probably just cheaper to let the gpu do it than having the cpu tied up doing it on the dc aggressively .

    Radirgy precious ps2 : The game clips extremely close to the view port,to prevent wasted polys. I guess fillrate wasnt an issue this time around as backface culling wasnt used though.
    3825 tris per frame 60 fps = 229K pps.


    As you can see the mesh mirrors very closely exactly what the ps2 is drawing, only what is seen on screen.


    Here we look at the underside of ships, they arent clipped so we can be sure backface culling wasnt used on them:



    Radirgy Dreamcast: On the dreamcast it draws way beyond the view port, inflating the polygon count. Doesnt mean the cpu isnt calculating them, it just means alot is unseen and quite unnecessary .

    30,612 tris per frame 60fps = 1.8 mpps.


    Way more stuff being rendered than seen:


    Lets zoom in on the DC dumped mesh to the section to try and see where would the ps2 be rendering, as you can see its only a extremely small part of what was extracted on the dc.

  11. #2591
    Raging in the Streets Yharnamresident's Avatar
    Join Date
    May 2013
    Location
    British Columbia
    Posts
    3,769
    Rep Power
    62

    Default

    Quote Originally Posted by Cloofoofoo View Post
    Another comparison , Radirgy ps2 vs dreamcast. Both games sport the same assets BUT rendering behavior is extremely different. On the ps2 the game does not backface cull the ships but it does view frustum clipping. Meaning unseen meshes will be tested for visibility against the camera and discarded if tested not visible.On the dc they did no such thing because of how the gpu works and how simply it doesnt draw beyond the viewports and most likely draws less than what the cpu sends. BUT the cpu is still calculating those models outside of view because they didnt manually clip before sending to the gpu, so technically alot shmups have inflated polygons count on the dc because whats not seen is being shows to us once we extract. I suspect the reason for this its because its probably just cheaper to let the gpu do it than having the cpu tied up doing it on the dc aggressively .
    I get what you're saying, but it seems impossible for them to clip games to the point where they have N64 polycounts. Like you said Space Channel 5: Part 2 was heavily clipped, that game was still around 27,000 polygons.

    Also nice Soul Calibur comparison, can't rep.

    The funny thing about Soul Calibur, is clearly Namco was testing the Dreamcast waters with that game. It ended up selling 1 million copies, yet they still didn't create other Dreamcast projects.
    Certified F-Zero GX fanboy

  12. #2592
    Hedgehog-in-Training Hedgehog-in-TrainingWildside Expert
    Join Date
    Aug 2018
    Posts
    243
    Rep Power
    6

    Default

    Quote Originally Posted by Yharnamresident View Post
    I get what you're saying, but it seems impossible for them to clip games to the point where they have N64 polycounts. Like you said Space Channel 5: Part 2 was heavily clipped, that game was still around 27,000 polygons.

    Also nice Soul Calibur comparison, can't rep.

    The funny thing about Soul Calibur, is clearly Namco was testing the Dreamcast waters with that game. It ended up selling 1 million copies, yet they still didn't create other Dreamcast projects.
    Different usage, basically shmups has things all laid out. So alot of the models lay out of view which means this is 100% ideal for view frustum clipping. Even in other situations like resident evil 4 were totaly unclipped = 65K tris in the frame on pc to ps2 view port clipping + backface culling on the stage = 12K to 25K per frame using the same assets,see massive reduction. Basically think about it like this if the game has more stuff on screen fully or partially then view port clipping is less effective. For the shmups you got whole chunks of lands and enemies offscreen , it would be wasteful to not clip it. On the ps2 you could dedicate one of the co processors to this so its probably much cheaper for it to clip. So it seems incredible but in these specific shmups it lent itself extreme clipping like that(hello triggerheart) less so on resident evil 4 but still effective. I might be simplifying because they might have to do otherthings like dividing meshes on the fly for more efficient clipping.

    We would need to check space channel 5 on the ps2 to see how it operates but on the dreamcast they seemed less concerned with what was offscreen instead they opted for the rarely used backface culling on the dc which means they probably wanted to prevent overdraw/fillrate problems. To be clear we may be extracting these polygons but the dreamcast could most likely be rendering even less due to how the gpu operates regardless whats being sent to it.

    Edit: they made more games on naomi, look at mazan from namco. They use nice framebuffer effects and figured out a cheap trick for refraction that looks convincing.
    Edit2: heres a image to help you visualize view frustum clipping.

    Last edited by Cloofoofoo; 12-07-2019 at 03:49 PM.

  13. #2593
    Raging in the Streets Yharnamresident's Avatar
    Join Date
    May 2013
    Location
    British Columbia
    Posts
    3,769
    Rep Power
    62

    Default

    Quote Originally Posted by Cloofoofoo View Post
    Different usage, basically shmups has things all laid out. So alot of the models lay out of view which means this is 100% ideal for view frustum clipping. Even in other situations like resident evil 4 were totaly unclipped = 65K tris in the frame on pc to ps2 view port clipping + backface culling on the stage = 12K to 25K per frame using the same assets,see massive reduction. Basically think about it like this if the game has more stuff on screen fully or partially then view port clipping is less effective. For the shmups you got whole chunks of lands and enemies offscreen , it would be wasteful to not clip it. On the ps2 you could dedicate one of the co processors to this so its probably much cheaper for it to clip. So it seems incredible but in these specific shmups it lent itself extreme clipping like that(hello triggerheart) less so on resident evil 4 but still effective. I might be simplifying because they might have to do otherthings like dividing meshes on the fly for more efficient clipping.
    What you said about Resident Evil 4 seems much more convincing and realistic.


    But we're talking going from 48,000 polygons:




    to 6800 polygons:




    In the Dreamcast version you can see the game does indeed render polygons behind the black bars. But you can still see its super dense in polygons.


    If we look at Chaos Field, the polycounts are slightly lower in the PS2 version because they likely did do some clipping, but its not 1/6th the polycounts:



    22,000 polygons




    24,500 polygons

    No surprise the wireframe is fine:




    Also the game can spike up pretty high. This is when I was taking sloppy screenshots and didn't zoom in, so you can 90% of the polygons are located in the viewing window:




    37,000 polygons


    Don't worry Cloofoofoo I am taking into consideration what you are saying. But I've been using ZeroGS for a year and a half and almost literally everytime the wireframe is messed-up, the plug-in gives N64 polycounts.
    Certified F-Zero GX fanboy

  14. #2594
    Raging in the Streets Yharnamresident's Avatar
    Join Date
    May 2013
    Location
    British Columbia
    Posts
    3,769
    Rep Power
    62

    Default

    Also I'm working on a Ratchet and Clank post.
    Certified F-Zero GX fanboy

  15. #2595
    Raging in the Streets Yharnamresident's Avatar
    Join Date
    May 2013
    Location
    British Columbia
    Posts
    3,769
    Rep Power
    62

    Default

    Ratchet and Clank is one of the first games to utilize the PS2's strengths. I can imagine a lot of people finally started seeing what the PS2 is capable of once they saw this game. Unlike a lot of posts in this thread, this isn't a "could it be ported to Dreamcast" post, this is one of the first PS2 games to take a far leap beyond Dreamcast capabilities.


    But first we gotta look at this PS2 Performance Analyzer post:





    It says 6.2 million PPS.

    I went to the same spot in ZeroGS and it gave me 4.9 million PPS. This isn't N64 polycounts(which is usually the case), but its still not the same and I will believe the Performance Analyzer more than ZeroGS.



    Of course the wireframe is a glitchy mess, enemies painting the screen:



    So yes I don't feel I'm getting accurate polycounts. ZeroGS is likely giving me 80% of the real polycounts.


    This is to SegaAMD. You can see this game is using 8-colour textures, look at the path texture:



    Or these mountains:



    A very low amount of colours, its consistent through this entire game.


    Heres the second stage. This screenshot is likely around 110,000 polygons:



    I'd estimate this game hovers around 3.5 to 7 million PPS. I got the game on my game shelve, so if I find a way to get absolutely accurate polycounts, then I will thoroughly go through it.
    Certified F-Zero GX fanboy

Thread Information

Users Browsing this Thread

There are currently 5 users browsing this thread. (1 members and 4 guests)

  1. Cloofoofoo

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •