This forum uses cookies
This forum makes use of cookies to store your login information if you are registered, and your last visit if you are not. Cookies are small text documents stored on your computer; the cookies set by this forum can only be used on this website and pose no security risk. Cookies on this forum also track the specific topics you have read and when you last read them. Please confirm whether you accept or reject these cookies being set.

A cookie will be stored in your browser regardless of choice to prevent you being asked this question again. You will be able to change your cookie settings at any time using the link in the footer.

Thread Rating:
  • 8 Vote(s) - 3.75 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Final Fantasy IV Complete Collection - ULUS10560
Hi everybody.

I need help with JPCSP and ISO Final Fantasy IV CC.

Im emulated correct it but in battle i see "field" black.

See in IMG:

http://img198.imageshack.us/img198/9818/sinttulorvu.jpg


And another problem, animation of skills dont see correct:

http://img687.imageshack.us/img687/2118/...lo1flt.jpg

Can help me please?

Reply
(07-04-2012, 01:10 PM)MadCabbit Wrote:
(07-04-2012, 05:41 AM)hyakki Wrote: how exactly did you comment out re.enableFlag(GU_STENCIL_TEST); ? did you recompile it from source?..

your best bet for now its just to test on revision r2535 enable the settings like these, if that works you can attempt to modify the newer revisions with the change.

Code:
|Video|
[ ] Disable VBO (automatically disabled if not supported)
[ ] Only GE graphics (not recommended for homebrew)
[ ] Use vertex cache
[x] Use shaders^experimental
[x] Use a Geometry Shader for 2D rendering (only relevant when using shaders)
[ ] Disable UBO (automatically disabled if not supported, only relevant when using shaders)
[ ] Enable VAO (only relevant when using shaders and vertex cache)
[ ] Enable saving GE screen to Textures instead of Memory
[ ] Enable decoding of indexed textures (using CLUT) in shader (only relevant when using shaders)
[ ] Enable dynamic shader generation (only relevant when using shaders, recommended for AMD/ATI)
[x] Enable the shader implementation for the 'Stencil Test' (more accurate but slower)
[ ] Enable the shader implementation for the 'Color Mask' (more accurate but slower)
[ ] Disable optimized Vertexlnfo reading (may improve compatibility)
[ ] Use Software Rendering (very experimental, not yet complete, slow)

|Memory|
[] Ignore invalid memory access (not recommended for test reports, but improves performance)!
[ ] Ignore unmapped imports (not recommended)

|Media|
[x] Use Media Engind
[ ] Extract media files to tmp folder
[x] Decode audio files with SonicStage (must be installed separately and requires Media Engine)
[ ] Use debug font (disable flashO fonts)

I changed
re.enableFlag(GU_STENCIL_TEST);
to
/* re.enableFlag(GU_STENCIL_TEST); */

I tried r2535 too, and had problems; I was only able to get 2490 working. I was using the settings in the first post, though. Using these settings, r2535 is working with the Stencil Test, and so is the latest, but it has the missing font bug. Is there any fix for that, or is it known what is the last revision that doesn't have this bug?


I was wondering if someone could give me a little insight on how to make the 're.enableFlag(GU_STENCIL_TEST)' change. I've tried everything I can think of but am not that fluent in Java. Any help is appreciated!
Reply
your current best bet is to just use r2491 since you can use 'save ge to textues' along with the 'stencil test' options to give the game a nice speedup (esp noticeable in the boss battles), and you wont have to modify the source.
Reply
(09-12-2012, 10:40 AM)hyakki Wrote: your current best bet is to just use r2491 since you can use 'save ge to textues' along with the 'stencil test' options to give the game a nice speedup (esp noticeable in the boss battles), and you wont have to modify the source.

is there a link for r2491? it says there is no build on the Automated Jpcsp builds site. I am trying to get text to show up. everything else seems to work great besides that.
Reply
(09-12-2012, 10:40 AM)hyakki Wrote: your current best bet is to just use r2491 since you can use 'save ge to textues' along with the 'stencil test' options to give the game a nice speedup (esp noticeable in the boss battles), and you wont have to modify the source.
how to make the 're.enableFlag(GU_STENCIL_TEST)' change ? or can you please post last revision with 're.enableFlag(GU_STENCIL_TEST)' change ?
Reply
r2733 displays the text if Enable shader implementation for the "Color Mask" is checked. At least for me it does.
Reply
(09-23-2012, 04:39 PM)latewinterstorm Wrote: r2733 displays the text if Enable shader implementation for the "Color Mask" is checked. At least for me it does.

Yeah that's working again the only drawback to using color mask is it has semi transparent command menus on the battle screen, and the boss battle loading effect takes 5 min to load (unless you quickly toggle on Save GE to textures, then off again before the battle starts)
       

log and profiler from r2736 attached maybe it can help discover new information on the boss loading screens (gpu usage is low on this scene)

.zip   Log_r2736(64bit)_INFO.zip (Size: 143.06 KB / Downloads: 173)

I did another test with software rendering game looks and runs good except on some scenes only the left screen side is displayed correctly (battle loading screen, battle backgrounds)
Software Rendering
           

.zip   Log_r2736(64bit)_INFO(software_render).zip (Size: 142 KB / Downloads: 152)




off-topic
I found that if you increase the ringbuffer in scepsmfplayer.java you can prevent the movie audio from skipping (though it will still be about 3 seconds desynced by the movies end (the movie plays about 1~2fps slow at 28 fps instead of 29.97))
Code:
//change at own risk , I assume this will break proper psp emulation
psmfMaxAheadTimestamp = sceMpeg.getMaxAheadTimestamp(581 * 2);
Reply
(09-24-2012, 09:41 PM)hyakki Wrote: log and profiler from r2736 attached maybe it can help discover new information on the boss loading screens (gpu usage is low on this scene)
The best would be to "Reset the Profiler Information" just before the start of the loading screen and quit Jpcsp just after it. This would capture profiler information just for this slow loading sequence.

From the profiler info you provided, it seems that only a few methods are used intensively:
Code:
_S1_2_88F8DD0 847,509,440 instructions (15.357%), 103,335 calls (088F8DD0 - 088F8E8C, length 48)
_S1_2_8A062F0 788,938,191 instructions (14.296%), 275,949 calls (08A062F0 - 08A06754, length 282)
_S1_2_88CD178 510,032,684 instructions (9.242%), 8,098 calls (088CD178 - 088CF584, length 2308)
_S1_2_88F87F8 482,409,533 instructions (8.742%), 69 calls (088F87F8 - 088F89D4, length 120)
It would be interesting to find out which ones are exactly used during the slow part to focus on them.

From a first analysis, _S1_2_88F8DD0 looks like a simple Bubble Sort algorithm and could be implemented into native Java.

Thanks!
Always include a complete log file at INFO level in your reports. Thanks! How to post a log
Reply
(09-25-2012, 05:16 PM)gid15 Wrote:
(09-24-2012, 09:41 PM)hyakki Wrote: log and profiler from r2736 attached maybe it can help discover new information on the boss loading screens (gpu usage is low on this scene)
The best would be to "Reset the Profiler Information" just before the start of the loading screen and quit Jpcsp just after it. This would capture profiler information just for this slow loading sequence.

From the profiler info you provided, it seems that only a few methods are used intensively:
Code:
_S1_2_88F8DD0 847,509,440 instructions (15.357%), 103,335 calls (088F8DD0 - 088F8E8C, length 48)
_S1_2_8A062F0 788,938,191 instructions (14.296%), 275,949 calls (08A062F0 - 08A06754, length 282)
_S1_2_88CD178 510,032,684 instructions (9.242%), 8,098 calls (088CD178 - 088CF584, length 2308)
_S1_2_88F87F8 482,409,533 instructions (8.742%), 69 calls (088F87F8 - 088F89D4, length 120)
It would be interesting to find out which ones are exactly used during the slow part to focus on them.

From a first analysis, _S1_2_88F8DD0 looks like a simple Bubble Sort algorithm and could be implemented into native Java.

Thanks!

here is a new profiler.txt, I reset it right before the battle screen and closed the emulator right after
Edit:
I added another profiler from another boss scene it seems to be slightly different instead of _S1_2_88F8DD0 , _S1_2_8A20374 was higher % this time)

(btw when Enable saving GE screen to Textures instead of Memory the screen loads 50% faster (still slow though), but turning that on breaks the color mask / stencil test option needed for the text :p)


Attached Files
.zip   Log_r2736(64bit)_INFO.zip (Size: 97.39 KB / Downloads: 149)
.zip   Log_r2736(64bit)_INFO(2nd boss).zip (Size: 91.68 KB / Downloads: 159)
Reply
Some new findings with Save GE to textures and this game.
Do this at own risk, this modification should only be used with this game and could cause problems with other games.

Code:
This will make Save GE to Textures compatible with color mask (half screen bug),
it also fixes numerous other problems in this game (Hit points are shown in battle again, slightly speeds up some effects like boss battle screen or some magic effects, that said you will also loose some effect like the battle swirl effect :p.. but I don't think that's even related to this)
---
VideoEngine.java
around line 4779 @ r2740 (or search)

geTexture = GETextureManager.getInstance().getGEResizedTexture(re, geTexture, tex_addr, bufferWidth, width, height, pixelFormat);

Change to

if (bufferWidth != 512)  {
geTexture = GETextureManager.getInstance().getGEResizedTexture(re, geTexture, tex_addr, bufferWidth, width, height, pixelFormat);    
}

# this is just a quick way I found to do it without re-writing to much code or removing it completely. commenting it out also works .
---
Reply


Forum Jump:


Users browsing this thread: 65 Guest(s)