PlayStation Development Network

I've been re-reading some old issues of EDGE Magazine (UK), and – according to reader correspondence in issues 67 & 68 from 1999 (see link below) – some people were claiming that newer models of the PlayStation / PS1 (presumably the SCPH-5552 or SCPH-7502 ?) featured slightly better graphics than the original SCPH-1002 model (specific games mentioned were Micro Machines and Soul Blade).

Does anybody know anything about this? I remember reading complaints from 1994 that the PS1 had certain 'bugs' in the way that it handled polygons, so were these rectified in later models?

I know that the newer PlayStations lacked the RCA phono outputs and therefore a different RF lead was provided instead (which connected to the A/V multi out socket) but surely this wasn't the only cause of the 'improved' graphics?

Link: http://read.oldgamemags.com/Multi-forma ... ions/Edge/

I think you're referring to the hardware change which was from VRAM to SGRAM.
This is why Sony made two debug consoles. The blue and jade green. Both had either chipset to test their games with.

As for concrete results if VRAM over SGRAM causes any difference, I don't know.

[youtube]http://www.youtube.com/watch?v=_HKjcxCuRfM[/youtube]

I recall SCPH-750x and later systems have a much sharper video output than the SCPH-700x and possibly older units as they have a slightly blurred output even with S-Video... I think that's the slight graphical improvement they're talking about.

interesting ! looking at the video, I think it's more about dithering
I have had a similar problem with my RPG game, when I draw a gradient rectangle (GT4) sometimes the dithering doesn't work.
I activated the dithering in the GsInitGraph function, but some times, depending on how much sprites are shown on screen, the dithering on a gradient polygon won't work ...
It seems that this bug appears when I use the GsSortFixBg16 function, but it also appear on the PSone model.

Are 5501 affected too?

I mean, to be 100% confirmed?

Thanks for the JPG, that's making the difference much more obivous as in the movie. It looks like lower color depth... or missing texture... though the right image doesn't doesn't seem to have a texture pattern either... but maybe that's just because "texture+dither" is producing less dominant dithering than "raw dither" without texture.

I don't really know what happened there & why. Some ideas...
- old GPUs didn't support dithering at all ???
- dither enable/disable works slightly different on old/new GPUs?
- dithering takes more clock cycles on old GPUs (?) and the game has intententionally disabled dither to gain more rendering time?

Btw. which game/scene is that screenshot/movie from exactly?

nocash wrote:Btw. which game/scene is that screenshot/movie from exactly?

It's the first level of the first tomb raider game.
by the way, does controller #2 work on no$psx?

The old GPU only use 5 bits when modulating the texture color while the new GPU uses all 8 bits. That might explain the banding. You can test the difference with my GPU plugin. It's also documented on my wiki.

I've been running some GPU tests on several boards this morning:
EARLY-PU-8 board (with 160pin GPU "CXD8514Q") (SCPH-1002) (dated around 1995)
LATE-PU-8 board (with 208pin GPU "CXD8561Q") (SCPH-1002) (dated around 1996)
PM-41 board (with 208pin GPU "CXD8561CQ") (SCPH-102, PSone) (dated around 1999)

Raw gouraud shaded polygons (without texture) are dithered smoothly on all GPUs.

Textured gouraud shaded polygons are dithered too, but, as AmiDog mentioned, the texture blending on old GPU drops the lower 3bit of the gouraud shaded color, so each 8 shading levels are forcefully getting the same dithering applied, and then the next higher 8 shading levels are abruptly getting higher brightness. That effect occurs only on the old 160pin GPU, the later 208pin GPUs on LATE-PU-8 and PM-41 boards are having smoother dithering, with different dithering applied to each shading level.
I've tried to take photos of my tests, but photographing the TV screen didn't work out out too well. Anyways, the test results are looking pretty identical to the "banding" & "no banding" pictures posted above. So, I am quite sure that I've reproduced the "banding" glitch... and, it's been apparently fixed on LATE-PU-8 boards, ie. about 1996...

yaroze wrote:I've been re-reading some old issues of EDGE Magazine (UK), and – according to reader correspondence in issues 67 & 68 from 1999 (see link below) – some people were claiming that newer models of the PlayStation / PS1 (presumably the SCPH-5552 or SCPH-7502 ?) featured slightly better graphics than the original SCPH-1002 model (specific games mentioned were Micro Machines and Soul Blade).

If the articles did actually refer to the "banding" issue, then it must have been fixed around 1996 (or maybe late 1995), and model-wise, EARLY/LATE-PU-8 were both called SCPH-1002 in europe (or SCPH-1001 in us). So the change occurred between "original SCPH-1002" models and "later SCPH-1002" models. Only, japanese PSX'es seem to have a special model number (SCPH-5000 for consoles with LATE-PU-8 boards).

I am not too tempted to download the 200MB magazine scans - did somebody download them, and could quote that GPU related "reader correspondence" stuff? Or upload that section(s) as jpg? Just in case they've been talking about something entirely differently, like sharper picture, or richer colors, or whatever.
Oh, and screenshots for those "Micro Machines and Soul Blade" games on old/new GPUs would be nice, too!

[youtube]http://www.youtube.com/watch?v=jeY0atkyT_c[/youtube]
Curiously, the banding seen in the software rendered DOS version of Tomb Raider at 4:21 here seems to correspond to the behavior on the old GPU and VRAM, while the smoother shadows seen on the new GPU and SGRAM look more similar to the hardware accelerated PC versions.

Since the game came out in late '96, maybe they fixed this on purpose for the newer silicon?

Does the banding issue only affect the earlier 1001/1002 models which use PU7/PU8 boards? I'm thinking of picking up a 5501 (which is a PU-18 I think?). I just wanna make sure I'm making the right choice.

For completeness, I've made a texture shading comparison image with screens taken from my DTL-H2000 (which had the old GPU) and SCPH-7000 respectively. I captured the screenshots via framebuffer reads for maximum clarity. Shading is from maximum brightness (255) to black (0).



And yup, texture shading on the old GPU does indeed look a bit crappy compared to the new GPU.

This is my first post on this forum. Since quite a time I have been dealing with the use of playstations as CD player. So far my findings are following:
There are substantial differences between the various versions of the PSX, i.e. SCPH1002, SCPH5502, SCPH7502. There are more (SCPH7002, 9002,…) but I did not examine and test them.
Types 1002 and 5502 are using DACs of type AK4309 from Asahi-Kasei (discontinued delta-sigma DAC), whereas 1002 has got an OPAMP following the DAC the 5502 don`t. 7502 is built completely different, the audio signal just comes out of one of these big chips (I assume a DSP) and is fed to an I/V converter.
---------------------------------------------------------------------------------------------------------------------------------
Android App Development Company, Game Development Company

... and this relates to this topic and the GPU how exactly?

So the SCPH 7502 containts the "nonbanding" one right? (yay I am so lucky )

My guess is that >90% of consoles on the market have the newer chipset.
You have to be looking really hard for an early model.

Medis wrote:So the SCPH 7502 containts the "nonbanding" one right? (yay I am so lucky )

Yes. The only consoles that have the old GPU/VRAM are SCPH-1000, SCPH-3000 and early SCPH-1001 and SCPH-1002

SCPH-5000 and later SCPH-1001 and SCPH-1002 have the new GPU/SGRAM as do all the later consoles.

The blue cased debug consoles have the old GPU, the green ones have the new GPU. Yarozes have the new GPU, too.

LameGuy64 wrote: ↑November 13th, 2016, 4:07 pm For completeness, I've made a texture shading comparison image with screens taken from my DTL-H2000 (which had the old GPU) and SCPH-7000 respectively. I captured the screenshots via framebuffer reads for maximum clarity. Shading is from maximum brightness (255) to black (0).

And yup, texture shading on the old GPU does indeed look a bit crappy compared to the new GPU.

Is it possible to run this test using the PS2 backwards compatible PS1 mode? I'm currently investigating reports that PS1 on PS2 has different dithering?

Here are images from the two issues of EDGE in the OP. It does mention banding.

https://retrocdn.net/index.php?title=Fi ... f&page=136 https://retrocdn.net/index.php?title=Fi ... f&page=137 https://retrocdn.net/index.php?title=Fi ... f&page=129

My test program should run no problem on a PS2 under PS1 backwards compatibility. But the trouble is dumping the framebuffer as a raw bitmap. There's no way of communicating to a PC in PS1 mode I believe so I guess saving to memory card and using compression would be the only option.