Open daniel229 opened 10 years ago
Which texture is it? Does it look like it ought to be rendered at all?
-[Unknown]
just compare to psp's,it does not ought to be rendered.
the ghost is more clearly seeing here
before
It seems to me that it's drawing some bloom, and it accidentally latches onto a framebuffer instead of an actual in-memory texture.
It must be nearby. I wonder how common y-offset framebuffers are vs. just x offset. Maybe we reduce the subarea max or use a ratio instead of hard limit. Could also check remaining height vs total height of texture.
-[Unknown]
How does ef12694c4b926c9619c24764bc369adbdc38af87 look?
-[Unknown]
Look better,still a little bright.
God of War,shadow problem come back.
God of War's problem isn't handled yet, if it was fixed before it was by accident.
-[Unknown]
I see.
Although, I think we can handle that with just viewport hacks... I think? Well, at least for non-3d... hmm. I wonder if it draws in throughmode.
In its case all the render textures start on the same line, which is probably the safest way to detect the offset. I doubt games are doing y offsetting, but then, who knows...
-[Unknown]
just before the bright.
Hmm. Fixed blending without a texture, what are the vertex colors? Is lighting enabled?
-[Unknown]
lighiting is enabled
vertex colors
Has this improved at all with the vram mirror attachments maybe?
Also, if you comment out the block starting with if (fbInfo.yOffset > MAX_SUBAREA_Y_OFFSET_SAFE && addr > 0x04110000) {
, does it fix it maybe?
-[Unknown]
No,it doesn't change anything.comment out that block not help either.
If you double click alpha blend mode right before it's doing it, and change the value to 000000aa
, is it still too bright? I just want to make sure this is where the problem is coming from.
If so, then the brightness is coming only from the drawing, which seems confusing because acab898be0b6075ef13b5f856753abb23dda8874 shouldn't affect that? Hmm.
-[Unknown]
too bright is gone
compare
Does this look correct in the software renderer?
-[Unknown]
I am not really sure if this how it should look but it looks somewhat right based on PSP screenshot earlier on There's also glitch when you leave the classroom that I am pretty sure is caused by camera being too high but it's probably not relevant
Could you try exporting a GE debugger dump on PC?
To do this, open the game and select Debug -> GE debugger..., then when it's displaying the scene, press Record in the top right. After a second or so, it'll finish and save a trace of the drawing activity.
After that, check the memstick/PSP/SYSTEM/DUMP folder and it'll have created a file named something like "ULES12345_0000.ppdmp". You can zip that and then drag and drop it into a reply here.
-[Unknown]
Shadow is incorrect position using vulkan backend.
On that area the game got a graphics regression. Both opengl and vulkan. fate extra.zip
It starts like this, which looks fine enough:
Then at 452/593, it reads from 0x08929ce0 (the address would be different in the game, but in any case, not VRAM), which looks like static. It's a 64x64 texture, but it's only drawing from 40x8 of it. This is drawn with an alpha test and src.a + ONE blend. It then blooms with the static, which blurs the static a bit.
If the static is skipped, it looks like this:
Because the static is in RAM, I haven't tried this on a PSP; whatever operation caused the static isn't a GE drawing operation specifically, but is either a bad block transfer, CPU, HLE, or other issue. I'm sure the dump would show the same static artifacts on a PSP.
The interesting thing was that I'd previously thought this was an issue with render-to-texture detection. Since the dump shows a RAM address, it can't be VRAM (it intentionally maps VRAM -> VRAM to capture framebuffer effects correctly.) My best guess is this is block transfer or missing memory copy function hook related... but I wonder what it's trying to draw here.
-[Unknown]
Since https://github.com/hrydgard/ppsspp/commit/acab898be0b6075ef13b5f856753abb23dda8874
before