Forgive me if I'm stupid, which is highly possible, but if you have to a framemeister that outputs to HDMI and that's digital I think?
-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
DF Retro: Metal Gear Solid 2 - A Kojima Masterpiece
- Thread starter Theorry
- Start date
Forgive me if I'm stupid, which is highly possible, but if you have to a framemeister that outputs to HDMI and that's digital I think?
Reckless Abandon
Member
Loved the little details in the game like this. Also, how you could shoot melons, when you looked up in the rain there would be drops on the screen, how you could get caught jacking it in a locker.
Just a great game. Top 5 all time for me.
We have been blessed.
Oh, that's not a stupid question but, unfortunately, it does not work.
I thought this too as my second DF project ever WAS Zone of the Enders 2 HD back in 2013. I thought I could just capture PS2 from a Framemeister and analyse it but I didn't yet understand the limitations of our software. Now I get it.
The only reason digital signal helps is that, well, it eliminates image noise and produce a reliably clean picture.
Analog cables introduce noise into the image - sometimes minor enough that you can't even see it by eye. To analyse this footage, we have a "compression" flag that can be used along with a number defining the tolerance. This basically overlooks minor image flaws in order to try and pick out unique frames. Analog imagery, without this flag, just comes up as 60fps because analog noise refreshes at that rate and changes the image just enough that it believes there is a new frame being displayed.
The compression feature overcomes this BUT it also makes it nearly impossible to pick-up torn frames since the tear detection algorithm is no longer able to determine where the image is breaking. So analog + tearing = no way.
The ONLY workaround here would be to manually go through the footage, frame by frame, and add a tear line for each torn frame. That would, of course, take hours. No thanks!
NXGamer
Member
A valid question, I cannot comment on DF software but capturing analogue signal for frame analysis is much,much harder than digital for sure, but fully possible as I have been doing it for a while in my videos.
It did take me sometime to write an algorithm that I felt worked enough and was confident in multiple tests to be very accurate. Tear tests on this are harder still but again possible and take much longer. It may be that I capture within my own software rather than use a 3Rd party that may help though.
Thanks for detailed posts, I guess the trade off with time consumption means its hardly worth it. Maybe. I never realised analog brought these problems but the potential of noise makes sense.
Great retrospective btw,
sixteen-bit
Member
probably, i'm talking about the PS2 releases
bigboss370
Member
This series brings me such joy. Always loved MGS2 and replayed it so many times
This post was a few days back, but this is most likely a CPU bottlneck. PCSX2 will only use 2 cores by default.
I don't think it does. I believe the player must have been bleeding anyway (which requires a bandage in-game) and it just so happens to look like the glass is cutting him. I thought the same thing for a moment.
Maybe someone can correct me though.
tomwhite2004
Member
no it was my gpu as my cpu wasn't close to maxing regardless of resolution (hardware mode), if it was a cpu bottleneck then i still would have had frame drops on native resolution which i didnt. should note that i was testing in the opening cut scene which seems to be far more demanding than in game but a good stress test.
I'll test the game real quick. I have a GTX 950 as well.
Ok, so at 3x the intro stays at 90 fps or higher(1.5x speed) most of the time and drops to a low of 75 when snake jumps down. Even the part with his camo turning off stays above 60 fps.
I am only using basic blending accuracy though, I don't know if this game requires a higher setting. But that feature mostly hits the CPU, not GPU.
I am only using basic blending accuracy though, I don't know if this game requires a higher setting. But that feature mostly hits the CPU, not GPU.
tomwhite2004
Member
switching to the opengl hardware backend did the trick, pretty much locked 60fps at 3x even with the blending accuracy at medium.
it seems that i have been using the d3d 11 backend all this time not knowing that it was that much slower. most of the games i have played to completion have been much less demanding or only accurate in software mode so never noticed. so thanks for the late response and mentioning the blending or i would have never known, going to enjoy those extra frames!
pantsattack
Member
What a great video. I loved the fly-though of the level geometry and around SS landing on the deck.
D3D is faster for me though.. probably because it's skipping effects, or not rendering them properly.
Are you using an old version of the emulator, or any non default settings?
tomwhite2004
Member
Thats odd, i went back to d3d to double check and I still see the same drops.
Im using a 1.5.0 dev build from about three months ago. And as far as non default settings I only use the basic safe "Recommended" speed hacks on the cpu side.
With d3d 11 I use "allow 8-bit textures" , basic mipmapping and full crc hack level. With open gl i use those same settings together with accurate date and medium blending. Unless of course a game requires a specific setting to be changed as per the wiki.
Was always under the impression that whilst opengl was more accurate it was slower as a result. I swear the more I try to understand this emulator and get optimal performance for my build the less i know!
On second thoughts maybe you're right. I can't remember whether Snake's life bar was so low that he was bleeding. I'll have to check the video again.
joeygreco1985
Member
Fuck yeah this was a great episode. I loved seeing all of the little tricks the dev team used. Very informative stuff.
MGS2 was way ahead of its time, and is still impressive to this day. I have this on my wall downstairs
MGS2 was way ahead of its time, and is still impressive to this day. I have this on my wall downstairs
Maybe performance was improved in a recent build. Normally D3D and OpenGL perform similarly unless you use high levels of accurate blending, or if you have an AMD GPU.
Interesting, so the UltraHDMI was used for the Goldeneye / Perfect Dark FPS tests? neat!
While that's technically true, people often overestimate the differences in what hw delivered in that gen as well.dark10x said:
But really - PS2 never received a high-end space game in its entire lifespan(and neither did XBox IIRC), so there is no frame of reference how far those could be pushed anywhere other than GC. I guess DC did get a StarLancer port, but I never saw how well that turned out.
Btw - a note about textures on XB version - you are correct conversion process was partly at fault - they re-compressed 8bit/4bit textures -> DXTC, which is what introduces the artifacts. IIRC MGS source art was authored as palette pixel-art(not the usual "compress textures to palette automatically" approach) and DXTC is just not great at handling such data. Fall back would be to use non-compressed textures (they certainly had memory to spare on XB) but given that port performance was already not great, that probably wasn't an option in the time given.
Tye The Czar
Member
As they showed in the video, I'm much more prone these days to being wowed by interactivity than just pretty graphics.