PS3 [Attempt] Restoring XMB UI of Firmware 1.00 - 2.70 to 4.84+

Status
Not open for further replies.
Think you mean this one?

upload_2020-4-7_3-55-43-png.25182


Tried making it all black:
View attachment 25183

No luck though :(

View attachment 25184
Im re-reading again this thread...

First thing that worths to be mentioned from the image amb.bmp is that is shocking that the XMB is not taking the colors from it because it seems to be designed for that exact purpose
The 12 vertical pixels are the 12 months of the year, and the 16 horizontal pixels are the 16 hours of sunlight of a day (the other 8 hours missing are the night)

I was trying to find the colors somewhere else (in other places different than amb.bmp), but the first thing required for this research was to know the exact color codes in hexadecimal, @Louay was helping me with this because he had a video capture card with nice quality (one of that elgato something) that it seems it was capturing the colors with high precission
We was also talking about the posible post processing filters of his TV, etc... trying to clean the screenshots the most posible, also he made several screenshots with different background colors just incase the little squares i was using as "samples" was merged with the background
The "samples" im talking about are the tiny squares in the theme settings, where can be seen the colors for the 12 months together in a single list, i was using the color picker in photoshop to get the hexadecimal value... and i was trying to find that same hexadecimal values inside some decrypted .sprx (trying to prove the theory that this colors could be hardcoded inside some sprx, but no luck with that)
There is some thread in the forum where we was doing this, i cant find it right now, but the resulting image was this, really interesting, keep it somewhere well stored because in my oppinion this is the most accurate we got to them

qi7tleI.png


Are 4 different screenshots overlapped, and the resulting hexcode next to them
As you can see there are some colors that was matching exactly in the 4 screenshots (this is a good signal it means the hexcode probably is exactly that one)
There are a few others where there is a tiny difference in the hexadecimal value in between the 4 screenshots (as example august)... still even this ones seems precise enought to give us a chance to find them

The theory is... all that colors are a lot of bytes (3 or 4 bytes * 12 months) that should be stored together
So incase the experiment we did to capture his hexadecimal codes is right... we have a lot of bytes to search for... we dont know exactly how could be ordered inside a sprx but for sure are going to be very close to the others... so is just a matter of finding one of them and looking around it to see if the are others matching
Sadly i could not find anything in the decrypted sprx i was looking at (not sure which ones it was)
 
Last edited:
Good finding :)
Yeah, if someone wants to continue this research take a read at what i was trying, just to dont repeat the same search in the same sprx files

The same search can be made in the files inside .qrc files (.mnu mostly), at the point i made this didnt existed Xforge... but right now i dont think that colors are loaded from the qrc files, dunno, this is a still a mistery unsolved

...if someone wants to continue this research take a read at what i was trying, just to dont repeat the same search in the same sprx files
I think i just realized how it works, and why my theory was partially wrong :/

The 12 tiny squares filled with colors that can be seen in the "side menu" of the [XMB]->[Settings]->[Theme Settings] with a color for every month of the year are 12 <plane> objects
Doesnt seems to exists inside any .rco file so it looks like there is some .sprx creating them on runtime so we can deduce that colors are hardcoded inside some .sprx (this part of the theory probably is correct)

The way how the color works in the <plane> objects have a value specific for every channel, and are data type "float", just to show you an example (line has been shorted keeping only the related attributes)
Code:
<Plane name="December" colorScaleR="0.1" colorScaleG="1" colorScaleB="2" colorScaleA="0.5"></Plane>

So is needed to decompose and convert the hexadecimal values i captured with the color picker in photoshop... to... 4 consecutive float values (not sure how is stored that inside an .sprx or even his endianess)
Not sure how to do it correctly right now, but thats the bone to follow, i can smell it :mad:
 
Last edited by a moderator:
I have been working on adding the tex_busy and tex_playing animated icons to imagefont.bin.

Quite a lot of work involved really
  • Extract original gim images from rcos
  • convert to pngs
  • cut out each of the 30 frames into separate images
  • convert to indexed palettes
  • use pngquant to force all frames to have same palette
  • rename to xxxxframe001 to xxxxframe030
  • create cfg file for the imagefont.bin tool
  • built imagefont.bin

Here is where i have got to. not looking too bad but not perfect either

I want to combine the shadow images into it too if possible, however I am not sure how to do this as the shadow images have a black background. I am not sure how to make them have proper transparency like they are meant to have. The system is converting the black to transparency in some way I do not fully understand. I have tried replacing black with transparent but its not right.

@sandungas maybe you understand how to convert these shadow images so they are on transparent backgrounds? My attempts so far have failed. I wanted to try make then properly transparent and then combine the top layer images with the shadow images so each frame is a single png with the shadow included. Any ideas

tex_playing_shadow.png
tex_busy_shadow.png



EDIT: OK, I have got it sorted, thanks a lot to Kono on discord for the help.
 
Last edited:
@Danxx444 I can't remember was it you who wanted to change the way the background is pink during the coldboot, and then it changes to the user set background when the XMB loads? :)

Cos I just figured it out thanks to that new tool Xforge. :)

View attachment 25178

View attachment 25179
I'm sorry it took so long... it's almost a year since you mentioned me lol...but that's amazing ! you can always discover the mysterious behind the xmb... good job brother.
 
I have been working on adding the tex_busy and tex_playing animated icons to imagefont.bin.

Quite a lot of work involved really
  • Extract original gim images from rcos
  • convert to pngs
  • cut out each of the 30 frames into separate images
  • convert to indexed palettes
  • use pngquant to force all frames to have same palette
  • rename to xxxxframe001 to xxxxframe030
  • create cfg file for the imagefont.bin tool
  • built imagefont.bin

Here is where i have got to. not looking too bad but not perfect either

I want to combine the shadow images into it too if possible, however I am not sure how to do this as the shadow images have a black background. I am not sure how to make them have proper transparency like they are meant to have. The system is converting the black to transparency in some way I do not fully understand. I have tried replacing black with transparent but its not right.

@sandungas maybe you understand how to convert these shadow images so they are on transparent backgrounds? My attempts so far have failed. I wanted to try make then properly transparent and then combine the top layer images with the shadow images so each frame is a single png with the shadow included. Any ideas

View attachment 32601 View attachment 32602


EDIT: OK, I have got it sorted, thanks a lot to Kono on discord for the help.
You managed to solve some of those xmb mysteries, which me and the @sandungas were working on to find out something.
I really wish I could test your new mods..
 
It's a shame I've been out all this time, I've had problems and now my PS3 doesn't work anymore. :(
That's a shame. Maybe you will be able to pick up another, I am finding PS3s very cheap since PS5 came out. I got a few recently for 30euro to 40euro.


I figured this out with help from Kono on discord to convert the shadow PNG to have proper transparency.

120 conversions later (60 images combined now to make each animation) and here we are.

Now these can be used anywhere in XMBML, SFO, notifications , rcos etc.


I will release the upgraded imagefont.bin soon
 
I figured this out with help from Kono on discord to convert the shadow PNG to have proper transparency.

120 conversions later (60 images combined now to make each animation) and here we are.

Now these can be used anywhere in XMBML, SFO, notifications , rcos etc.
What a wonderful discovery man :) !... So you mean that now will be possible to use this waterdrop instead of the analog clock that's in the corner of the screen?
 
What a wonderful discovery man :) !... So you mean that now will be possible to use this waterdrop instead of the analog clock that's in the corner of the screen?
Not sure about that, possibly not. I do not know much about the clock and how it switches from showing the time to being a loading indicator. I will release the upgraded bin soon and maybe you can take a look sometime.

BTW as well as adding the 2 animations I have been adding a load of new tags to imagefont.bin for use in XMB menus.

For example
U+F5F2_frame001.png U+F5F3_frame001.png U+F604 _frame001.png U+F605 _frame001.png U+F608 _frame001.png U+F609 _frame001.png U+F4C0_frame001.png U+F4C1_frame001.png U+F4CB_frame001.png U+F4CE_frame001.png U+F4CF_frame001.png U+F5AD_frame001.png U+F5D2_frame001.png U+F56B_frame001.png U+F455_frame001.png U+F536_frame001.png U+F54E_frame001.png U+F54F_frame001.png U+F4E1_frame001.png U+F4E2_frame001.png U+F4E3_frame001.png U+F5D5_frame001.png U+F5D6_frame001.png U+F5D7_frame001.png U+F5E3_frame001.png U+F5FA_frame001.png U+F45A_frame001.png

There are about 250 of them in total. :)

upload_2021-3-31_22-23-15.png upload_2021-3-31_22-23-49.png upload_2021-3-31_22-28-24.png upload_2021-3-31_22-29-56.png
 
I want to combine the shadow images into it too if possible
I was talking about a problem around page 10 of this thread, but i cant find it now, i forgot some details about it but i remember well what they did
At some point they decided to "invert" the order of the frames in the waterdrop animation, but they forgot to "invert" his shadows, the result is the shadows only matches at 2 points of the sequence (at the center and at the extremes). Visually looks a bit weird because there is a lot of black pixels (not covered by the white pixels), also it gives some feeling like the shadow is "disaligned" with the waterdrop effect

The solution is to crop the sequence of the shadows image and reorder the frames... but most probably is going to be even better to forget about that shadows and create them yourself
In photoshop, with the "layer effects" in the white pixels, use the "drop shadow" with a displacement of 3 pixels and spreading another 3 pixels or so... and 25% transparency

The other animation with the "spinning atom" in OFW is perfect, his quality is way higher than the waterdrop, and his shadows are fine, but maybe you could do the same, delete the shadows and create your owns
 
At some point they decided to "invert" the order of the frames in the waterdrop animation, but they forgot to "invert" his shadows, the result is the shadows only matches at 2 points of the sequence (at the center and at the extremes). Visually looks a bit weird because there is a lot of black pixels (not covered by the white pixels), also it gives some feeling like the shadow is "disaligned" with the waterdrop effect
Well feck, that is annoying. :) I will sort it.
 
Hmmm... did not know that on ps3 had some of these tags... I remember modifying them with @LuanTeles, when I replace these with white tags, just like on ps4. Did you make these? Btw, this it's a godd idea ;)
 
Hmmm... did not know that on ps3 had some of these tags... I remember modifying them with @LuanTeles, when I replace these with white tags, just like on ps4. Did you make these? Btw, this it's a godd idea ;)

They are a few normally in the explore_plugin_full.rco, but they are not usable in custom mods when inside the rco, we have no way to call them in SFOs or XMBML or rco language files. By adding them to imagefont.bin it means we can use them anywhere easily, and also add loads of new ones related to CFW.

Berion made some and I made some.
 
They are a few normally in the explore_plugin_full.rco, but they are not usable in custom mods when inside the rco, we have no way to call them in SFOs or XMBML or rco language files. By adding them to imagefont.bin it means we can use them anywhere easily, and also add loads of new ones related to CFW.

Berion made some and I made some.
Yes! There are several tags in this rco, I remember when I was modifying it. You guys did an amazing job with these tags, it looks like they were made by the designer himself who made the original tags. Amazing man!
I also noticed that you work a lot with the imagefont.bin. that's because with it, it's easier to make magic happen. :D
 
Last edited:
At some point they decided to "invert" the order of the frames in the waterdrop animation, but they forgot to "invert" his shadows, the result is the shadows only matches at 2 points of the sequence (at the center and at the extremes). Visually looks a bit weird because there is a lot of black pixels (not covered by the white pixels), also it gives some feeling like the shadow is "disaligned" with the waterdrop effect
TBH after looking at the shadow and the image I think its the right way. Any other way I put it looks worse.

upload_2021-4-1_1-27-9.png
 
TBH after looking at the shadow and the image I think its the right way. Any other way I put it looks worse.
There are (at least) 2 versions of the waterdrop animation (tex_busy), the inverted order of the frames i mentioned can be appreciated better if you compare them
Code:
-system_plugin217.rco        MD5:8C320A8899E386C6743EFDD5C8EA2A71
-tex_busy217.gim             MD5:AC292D7AF3BA478A87400AC40FDA7574 (29x750 pixels -ps3rgba8888)
-tex_busy_shadow217.gim      MD5:862F858E85BEA609FF8E9C78EE73101A (32x960 pixels -ps3rgba8888)


-system_plugin251.rco        MD5:DDA512B03746913572121F8C5757B561
-tex_busy251.gim             MD5:AC24824CF4BD2DB82AE2FE512802968A (32x960 pixels -ps3dxt5)
-tex_busy_shadow251.gim      MD5:3D45ED809E9A81774D3CF5B3481105ED (32x960 pixels -ps3dxt5)

tex_busy217.gim (left)
tex_busy251.gim (right)
NhzrroY.png
z5DZf1u.png


tex_busy217.gim have the highest quality, is the original animation made in a 3D app (using fluids dynamics), they "captured" the frames of it and saved as rgba8888 preserving the quality
tex_busy251.gim was generated by increasing the pixel size of the previous version (this reduced the quality), but also, they stored it as dxt5 (this reduced the quality even more)

The "inversion" of the frames order in between firmware 2.17 and 2.51 can be seen a lot better in this comparison, this is "composite" test i made while i was trying to "replicate" the procedure made by sony designers
There are 2 images of each version because one is the original, and the other "clone" was made by me in photoshop, as you can see i matched them, i made this as a proof for myself to see that i was doing exactly the same procedure than sony designers
MGM5jUy.jpg


And this are the shadows
As you can see the shadows are identical, the only difference is in 2.17 was stored as rgba8888 and in 2.51 as dxt5
tex_busy_shadow217.gim (left)
tex_busy_shadow251.gim (right)
MnagwUW.png
Oxs8yty.png



-------------------------------
*The images with highest quality are the ones in 2.17
*The images of 2.51 was generated by modifying the images from 2.17
*The image quality was reduced in 2.51
*In 2.51 they inverted the frame order of the white pixels, but they forgot to invert the order of the shadows
 
Status
Not open for further replies.
Back
Top