PS3 [Research] Modifying the Gaia Visualization (custom_render_plugin/earth.qrc)

[DeViL303]

I was working on splitting the images and converting them from png to jpg. I had a good start on it but then I started playing with the QRCs & haven't touched it

That's cool. Anyway Ideally it could handle multiple formats of 360 image.

There are lots of these cubemaps and skyboxes available online at 4096x3072 so I decided to do those. Ideally it could take flat map and covert it to cubemap too.

The cool thing about this bat I made is that it can take images at any stage. So if you remove the cubemap.jpg and just have cubemap_rotated.jpg it will work from there, if there is no cubemap rotated already, and just face0.jpg to face5.jpg then it will go from there, and if there no face*.jpgs and there is only 01 to 24.jpg then it will go from there.

Also its my first bat like this ever, so probably needs optimizing.

 

[pink1]

I think to start with I'll finish getting everything done for the QRCs and let you guys play with that while I work on getting everything for the images going.
That way we can get some testing done & get some feedback while we get everything else finished up.

If that doesn't sound good just let me know.
Did you check out the other cubemap site I sent you the link to?

 

[DeViL303]

I think to start with I'll finish getting everything done for the QRCs and let you guys play with that while I work on getting everything for the images going.
That way we can get some testing done & get some feedback while we get everything else finished up.

If that doesn't sound good just let me know.
Did you check out the other cubemap site I sent you the link to?

Sounds great, I will keep working on the images anyway.

I did look at that site but could not seem to get high resolution, also had to save each image individually.

See this one: https://matheowis.github.io/HDRI-to-CubeMap/

All in one, multiple options for downloads, 4096x3072 cubemap, separate sides, resolution options. check it out. The only thing its missing is jpg output, but that not a big deal.

Also it seems to not have the annoying pinch effect at the poles..Savage

 

[DeViL303]

Waterworld:

Seamless textures are cool. This is just this texture x24, really easy to do.






@pink1 it would be cool to have an option for that if possible. To just tile the same image. Also another for applying an image to a face of the cube. I guess it can be done with bats too. I might look into that.

 

[pink1]

That sounds great. I have so many ideas for that part. I've wasted most of the day looking at different things for the cubemaps and imagine tools.
I need to get back on track I feel like I'm really slacking today lol

 

[DeViL303]

That sounds great. I have so many ideas for that part. I've wasted most of the day looking at different things for the cubemaps and imagine tools.
I need to get back on track I feel like I'm really slacking today lol

Maybe some of this could be integrated somehow or could be part of the toolkit..idk.

http://spiralgraphics.biz/viewer/index.htm
http://spiralgraphics.biz/packs/browse.htm

840 seamless gtx textures included, these are perfect for tiling x 24 . This app is perfect for this too. Also can render effect maps so we can get proper specular images afaik

 

[DeViL303]

Actually for tiled images, we can remap all to the same jpg, this way we have lots of room to work with for specular etc.

Circuit board mod:

 

[pink1]

Maybe some of this could be integrated somehow or could be part of the toolkit..idk.

View attachment 23120

View attachment 23119

http://spiralgraphics.biz/viewer/index.htm
http://spiralgraphics.biz/packs/browse.htm

840 seamless gtx textures included, these are perfect for tiling x 24 . This app is perfect for this too. Also can render effect maps so we can get proper specular images afaik

View attachment 23121

I think maybe we should just recommend using this if in a toolkit if needed for now because I'm not sure I can get something like that working without a lot of work & breaking compatibility with older systems.

I'm afraid if I try to add too much at once I'll end up with a bunch of half done projects. After we get a working tool that does the basics that we set out to do we may try making something a little more fancy.

 

[DeViL303]

@pink1 Cool. That app is amazing, exactly what was needed.

If we can not make the file bigger at any point I have an idea. We remove all the cloud textures, and split the remaining space between ground and specular. This way we can have high quality ground textures and a decent quality specular. As of now I can not use the specular at all. Some of the images are only 1.4kb. No idea how Sony could compress their images so much, I can not come close even at 99% compression.

Also I have some ideas for some built in simple patches, for example there could be tickboxes for making all 24 specular images either all black (no reflections) or all white (full reflections).

 

[DeViL303]

This app is really great, you can change the seed, so it generates new textures and effect maps.

Check it out @Berion , I think you will like it.

The GTX files it uses are tiny and great quality. Vector based I guess.

840 attached. 4.70MB

 

[sandungas]

But are you going to do an app specifc for earth.qrc or generic for all .qrc files ?
Im asking because looks like 2 different goals, most of the basic features as extracting or injecting files are going to work with lines.qrc icontex.qrc etc...


Edit:
From the tools we have been discussing before, i think jpegtran is one of the bests, i been reading about it and it looks all the transformations availables in the tool are losless it process the pixel info in a very special way, actually the tool doesnt allows to do any lossy transformation, this is why is not allowed basic functions like resize, lol
Anyway... is a little gem, and the source code is available, if you decide to add some "extended features" in my oppinion jpegtran could be a nice addition, and could be implemented in a generic way (not only for earth.qrc but for the other qrc files too incase they haves .jpg files)

It would allow to add a checkbox with the option "extract jpg files rotated 180" (and same for importing too)

 

[DeViL303]

Yeah its great. Good find.

I also found jpegoptim which is great as it allows to compress to specific filesize whic is just what we need.

jpegoptim --size=60k "Tiles\optimized\01.jpg"
jpegoptim --size=50k "Tiles\optimized\02.jpg"
jpegoptim --size=47k "Tiles\optimized\03.jpg"
jpegoptim --size=50k "Tiles\optimized\04.jpg"
jpegoptim --size=50k "Tiles\optimized\05.jpg"
jpegoptim --size=41k "Tiles\optimized\06.jpg"
jpegoptim --size=45k "Tiles\optimized\07.jpg"
jpegoptim --size=47k "Tiles\optimized\08.jpg"
jpegoptim --size=61k "Tiles\optimized\09.jpg"
jpegoptim --size=45k "Tiles\optimized\10.jpg"
jpegoptim --size=43k "Tiles\optimized\11.jpg"
jpegoptim --size=37k "Tiles\optimized\12.jpg"
jpegoptim --size=51k "Tiles\optimized\13.jpg"
jpegoptim --size=54k "Tiles\optimized\14.jpg"
jpegoptim --size=47k "Tiles\optimized\15.jpg"
jpegoptim --size=52k "Tiles\optimized\16.jpg"
jpegoptim --size=60k "Tiles\optimized\17.jpg"
jpegoptim --size=55k "Tiles\optimized\18.jpg"
jpegoptim --size=54k "Tiles\optimized\19.jpg"
jpegoptim --size=45k "Tiles\optimized\20.jpg"
jpegoptim --size=54k "Tiles\optimized\21.jpg"
jpegoptim --size=48k "Tiles\optimized\22.jpg"
jpegoptim --size=52k "Tiles\optimized\23.jpg"
jpegoptim --size=53k "Tiles\optimized\24.jpg"

I had to use sizes slightly smaller as its not exact compression, its closest guess I think. So its only 95% accurate approx.

 

[sandungas]

I been playing around with CubeTheSphere http://realmofconcepts.com/unreal/CubeTheSphere.html
I like it because doesnt runs in the web browser, is a tiny tool to do a single task and works fine
The only annoyance i found (and i fixed) from it is something that doesnt matters in most cases, but i will explain it later

It converts "sphere maps" into "cube maps"... the sphere maps are a rectangle of width=2*height
Im going to explain how to create the images in the same way are made in the official firmware
And im doing all this process with images of the earth, taken from https://visibleearth.nasa.gov/collection/1484/blue-marble?page=1

-----------------------------------------------
In the official PS3 images the equator have a lenght of 8 textures at 512 pixels each so... we have 4096 pixels in the equator
Is better to prepare the image with that dimmensions before processing it with CubeTheSphere
The point is... you can use bigger sizes, is ok, but the tool is going to take more time to render the cube faces
For that resize task i found a command line tool, and i realized is not so easy because most requires an installation, personally i always have installed irfanview but i wanted this to be more generic, so the only one i found is this one, it have an interface but also works in command line
http://www.rw-designer.com/picture-resize

The way it works is a bit special... you can pass the commands from a batch, but also you can rename the file with the settings and drop images on top of the exe
To do the resize into the correct format needed for CubeTheSphere download this file http://www.rw-designer.com/NT/6.0/PhotoResize400.exe
And rename it to PhotoResize4096x2048Q100O.exe
*4096x2048 is the final size
*Q100 is max quality
*O is silent mode

When using it in a .bat... whatever you write in command line overrides the settings from the filename, you can automatize it this way:

C:\>PhotoResizeWHATEVERNAME.exe -o -g100 -WxH4096x2048 test.jpg

----------------------------
After that the image is ready to be converted by CubeTheSphere
Export the images at 1024x1024
It creates 6 images named "front", "left", "right", "top", bottom", and "back"

In the PS3 the "cube face names" can be seen in the path... from 0 to 5, are:
0=back
1=left
2=front
3=right
4=bottom
5=top

You can rename the 6 images generated by CubeTeSphere like that
After that... is needed to automatize the actions with jpegtran like in the examples i mentioned in my previous posts to
1)Rotate them
2)Crop them
3)Rename them with the official filenames

 

[sandungas]

But what i explained in my previous post is not all.... here is the (small) annoyance i found from CubeTheSphere...
If you follow the steps i mentioned accuratelly... you are going to have valid images, ready to be injected in earth.qrc (visually everything is ok when you play it in the PS3)

The problem is when you compare the official images with the images resulting from what i explained... you are going to see that doesnt matches
And this is hard to explain why happens... im going to try it

In the oficial images we have the center of the front cube face that matches with the vertical line on top of madagascar
This is something sony decided to do this way (is the meridian 45º of the sphere), doesnt follows any standards

The program CubeTheSphere considers the center of the front cube face is located at the right border of the image (meridian 180º)

This sucks a bit... mostly because incase you are using a custom image made by you... probably you have the focus of interest in the center (what CubeTheSphere does by default is weird because considers the area at center of the image is back)
Anyway... the problem is obvious if you look at the NASA images and maps of the earth... in all them the meridian 0º is at the center

The solution is again jpegtran, try this bat to prepare the image for CubeTheSphere
This sequence of commands "cuts and pastes" sections of the image to locate the "front cube face" at the same position than the official images (and remember, all transformations made by jpegtrans are lossless, so this long and weird sequence of actions doesnt reduces his quality at all)

jpegtran.exe -crop 2560x2048+0+0 "test.jpg" "test_L.jpg"
jpegtran.exe -crop 1536x2048+2560+0 "test.jpg" "test_R.jpg"
jpegtran.exe -crop 4096x2048+0+0 "test_R.jpg" "test_RX.jpg"
jpegtran.exe -drop +1536+0 "test_L.jpg" "test_RX.jpg" "test_RL.jpg"

*Adapt the code to your needs... this is just an example i was doing with a "test.jpg" image (scaled to the correct dimmensions by the other app PhotoResize4096x2048Q100O.exe)

For reference use this image, the result should be exactly the same (but in better quality)

-------------------------------------
The resulting images

BACK (cube face 0)

LEFT (cube face 1)

FRONT (cube face 2)

RIGHT (cube face 3)

BOTTOM (cube face 4)

TOP (cube face 5)

 

[DeViL303]

Here is another bat, this one can also take a 512x512 "tile.jpg" or 1024x1024 "face.jpg" from Genetica Viewer or any other source and turn it into a "cubemap.jpg". The advantage to doing it this way is that all jpgs are optimized for their specific slot, rather than just make all jpgs 40KB to suit the smallest slot

This optional additional feature is only meant to be used on seamless tile-able images. You can still just use this like before with a "cubemap.jpg" if you already have one.

So goes from this:


To this:


To this:


Then it does the same as the last batch file by cutting the cubemap into 24 jpgs and optimizing it, then creating official file structure.

Spoiler: BAT

jpegtran -crop 1024x1024+0+0 -outfile face.jpg tile.jpg
jpegtran -drop +512+0 tile.jpg -outfile face.jpg face.jpg
jpegtran -drop +512+512 tile.jpg -outfile face.jpg face.jpg
jpegtran -drop +512+512 tile.jpg -outfile face.jpg face.jpg
jpegtran -drop +0+512 tile.jpg -outfile face.jpg face.jpg
jpegtran -crop 4096x3072+1024+0 -outfile cubemap.jpg face.jpg
jpegtran -drop +1024+1024 face.jpg -outfile cubemap.jpg cubemap.jpg
jpegtran -drop +1024+2048 face.jpg -outfile cubemap.jpg cubemap.jpg
jpegtran -drop +2048+1024 face.jpg -outfile cubemap.jpg cubemap.jpg
jpegtran -drop +3072+1024 face.jpg -outfile cubemap.jpg cubemap.jpg
jpegtran -drop +0+1024 face.jpg -outfile cubemap.jpg cubemap.jpg
if not exist "Cube_faces" MkDir "Cube_faces"
if not exist "Tiles\optimized" MkDir "Tiles\optimized"
for %%p in ("Cube_faces\cubemap.jpg") do (
jpegtran.exe -rotate 180 "cubemap.jpg" "Cube_faces\cubemap_rotated.jpg"
jpegtran.exe -crop 1024x1024+0+1024 "Cube_faces\cubemap_rotated.jpg" "Cube_faces\face0.jpg"
jpegtran.exe -crop 1024x1024+3072+1024 "Cube_faces\cubemap_rotated.jpg" "Cube_faces\face1.jpg"
jpegtran.exe -crop 1024x1024+2048+1024 "Cube_faces\cubemap_rotated.jpg" "Cube_faces\face2.jpg"
jpegtran.exe -crop 1024x1024+1024+1024 "Cube_faces\cubemap_rotated.jpg" "Cube_faces\face3.jpg"
jpegtran.exe -crop 1024x1024+2048+0 "Cube_faces\cubemap_rotated.jpg" "Cube_faces\face4.jpg"
jpegtran.exe -crop 1024x1024+2048+2048 "Cube_faces\cubemap_rotated.jpg" "Cube_faces\face5.jpg"
jpegtran.exe -crop 512x512+0+0 "Cube_faces\face0.jpg" "Tiles\01.jpg"
jpegtran.exe -crop 512x512+512+0 "Cube_faces\face0.jpg" "Tiles\02.jpg"
jpegtran.exe -crop 512x512+0+512 "Cube_faces\face0.jpg" "Tiles\03.jpg"
jpegtran.exe -crop 512x512+512+512 "Cube_faces\face0.jpg" "Tiles\04.jpg"
jpegtran.exe -crop 512x512+0+0 "Cube_faces\face1.jpg" "Tiles\05.jpg"
jpegtran.exe -crop 512x512+512+0 "Cube_faces\face1.jpg" "Tiles\06.jpg"
jpegtran.exe -crop 512x512+0+512 "Cube_faces\face1.jpg" "Tiles\07.jpg"
jpegtran.exe -crop 512x512+512+512 "Cube_faces\face1.jpg" "Tiles\08.jpg"
jpegtran.exe -crop 512x512+0+0 "Cube_faces\face2.jpg" "Tiles\09.jpg"
jpegtran.exe -crop 512x512+512+0 "Cube_faces\face2.jpg" "Tiles\10.jpg"
jpegtran.exe -crop 512x512+0+512 "Cube_faces\face2.jpg" "Tiles\11.jpg"
jpegtran.exe -crop 512x512+512+512 "Cube_faces\face2.jpg" "Tiles\12.jpg"
jpegtran.exe -crop 512x512+0+0 "Cube_faces\face3.jpg" "Tiles\13.jpg"
jpegtran.exe -crop 512x512+512+0 "Cube_faces\face3.jpg" "Tiles\14.jpg"
jpegtran.exe -crop 512x512+0+512 "Cube_faces\face3.jpg" "Tiles\15.jpg"
jpegtran.exe -crop 512x512+512+512 "Cube_faces\face3.jpg" "Tiles\16.jpg"
jpegtran.exe -crop 512x512+0+0 "Cube_faces\face4.jpg" "Tiles\17.jpg"
jpegtran.exe -crop 512x512+512+0 "Cube_faces\face4.jpg" "Tiles\18.jpg"
jpegtran.exe -crop 512x512+0+512 "Cube_faces\face4.jpg" "Tiles\19.jpg"
jpegtran.exe -crop 512x512+512+512 "Cube_faces\face4.jpg" "Tiles\20.jpg"
jpegtran.exe -crop 512x512+0+0 "Cube_faces\face5.jpg" "Tiles\21.jpg"
jpegtran.exe -crop 512x512+512+0 "Cube_faces\face5.jpg" "Tiles\22.jpg"
jpegtran.exe -crop 512x512+0+512 "Cube_faces\face5.jpg" "Tiles\23.jpg"
jpegtran.exe -crop 512x512+512+512 "Cube_faces\face5.jpg" "Tiles\24.jpg"
)
copy "Tiles\*.jpg" "Tiles\optimized\*.jpg"
jpegoptim --size=60k "Tiles\optimized\01.jpg"
jpegoptim --size=50k "Tiles\optimized\02.jpg"
jpegoptim --size=47k "Tiles\optimized\03.jpg"
jpegoptim --size=50k "Tiles\optimized\04.jpg"
jpegoptim --size=50k "Tiles\optimized\05.jpg"
jpegoptim --size=41k "Tiles\optimized\06.jpg"
jpegoptim --size=45k "Tiles\optimized\07.jpg"
jpegoptim --size=47k "Tiles\optimized\08.jpg"
jpegoptim --size=61k "Tiles\optimized\09.jpg"
jpegoptim --size=45k "Tiles\optimized\10.jpg"
jpegoptim --size=43k "Tiles\optimized\11.jpg"
jpegoptim --size=37k "Tiles\optimized\12.jpg"
jpegoptim --size=51k "Tiles\optimized\13.jpg"
jpegoptim --size=54k "Tiles\optimized\14.jpg"
jpegoptim --size=47k "Tiles\optimized\15.jpg"
jpegoptim --size=52k "Tiles\optimized\16.jpg"
jpegoptim --size=60k "Tiles\optimized\17.jpg"
jpegoptim --size=55k "Tiles\optimized\18.jpg"
jpegoptim --size=54k "Tiles\optimized\19.jpg"
jpegoptim --size=45k "Tiles\optimized\20.jpg"
jpegoptim --size=54k "Tiles\optimized\21.jpg"
jpegoptim --size=48k "Tiles\optimized\22.jpg"
jpegoptim --size=52k "Tiles\optimized\23.jpg"
jpegoptim --size=53k "Tiles\optimized\24.jpg"
if not exist "earth\flashrom\clouds\0\1\" MkDir "earth\flashrom\clouds\0\1\" && copy "Tiles\optimized\01.jpg" "earth\flashrom\clouds\0\1\000_000.jpg"
copy "Tiles\optimized\02.jpg" "earth\flashrom\clouds\0\1\000_001.jpg"
copy "Tiles\optimized\03.jpg" "earth\flashrom\clouds\0\1\001_000.jpg"
copy "Tiles\optimized\04.jpg" "earth\flashrom\clouds\0\1\001_001.jpg"
if not exist "earth\flashrom\clouds\1\1\" MkDir "earth\flashrom\clouds\1\1\" && copy "Tiles\optimized\05.jpg" "earth\flashrom\clouds\1\1\000_000.jpg"
copy "Tiles\optimized\06.jpg" "earth\flashrom\clouds\1\1\000_001.jpg"
copy "Tiles\optimized\07.jpg" "earth\flashrom\clouds\1\1\001_000.jpg"
copy "Tiles\optimized\08.jpg" "earth\flashrom\clouds\1\1\001_001.jpg"
if not exist "earth\flashrom\clouds\2\1\" MkDir "earth\flashrom\clouds\2\1\" && copy "Tiles\optimized\09.jpg" "earth\flashrom\clouds\2\1\000_000.jpg"
copy "Tiles\optimized\10.jpg" "earth\flashrom\clouds\2\1\000_001.jpg"
copy "Tiles\optimized\11.jpg" "earth\flashrom\clouds\2\1\001_000.jpg"
copy "Tiles\optimized\12.jpg" "earth\flashrom\clouds\2\1\001_001.jpg"
if not exist "earth\flashrom\clouds\3\1\" MkDir "earth\flashrom\clouds\3\1\" && copy "Tiles\optimized\13.jpg" "earth\flashrom\clouds\3\1\000_000.jpg"
copy "Tiles\optimized\14.jpg" "earth\flashrom\clouds\3\1\000_001.jpg"
copy "Tiles\optimized\15.jpg" "earth\flashrom\clouds\3\1\001_000.jpg"
copy "Tiles\optimized\16.jpg" "earth\flashrom\clouds\3\1\001_001.jpg"
if not exist "earth\flashrom\clouds\4\1\" MkDir "earth\flashrom\clouds\4\1\" && copy "Tiles\optimized\17.jpg" "earth\flashrom\clouds\4\1\000_000.jpg"
copy "Tiles\optimized\18.jpg" "earth\flashrom\clouds\4\1\000_001.jpg"
copy "Tiles\optimized\19.jpg" "earth\flashrom\clouds\4\1\001_000.jpg"
copy "Tiles\optimized\20.jpg" "earth\flashrom\clouds\4\1\001_001.jpg"
if not exist "earth\flashrom\clouds\5\1\" MkDir "earth\flashrom\clouds\5\1\" && copy "Tiles\optimized\21.jpg" "earth\flashrom\clouds\5\1\000_000.jpg"
copy "Tiles\optimized\22.jpg" "earth\flashrom\clouds\5\1\000_001.jpg"
copy "Tiles\optimized\23.jpg" "earth\flashrom\clouds\5\1\001_000.jpg"
copy "Tiles\optimized\24.jpg" "earth\flashrom\clouds\5\1\001_001.jpg"

• If there is no tile.jpg in the folder then it will look for face.jpg
• If there is no face.jpg then it will look for cubemap.jpg
• If there is no cubemap.jpg it will look for Cube_faces/cubemap_rotated.jpg
• If there is no Cube_faces/cubemap_rotated.jpg it will look for Cube_faces/face0.jpg to face5.jpg
• If there is no Cube_faces/face0 to face5 then it will look for Tiles/01.jpg to 24.jpg
• If there is no Tiles/01.jpg to Tiles/24/jpg it will look for Tiles/optimized/01.jpg to 24.jpg

 

[sandungas]

Take another look at my previous post @DeViL303, i updated it to add the images generated after the process
About what we was discussing... one of the most important things of that process i explained is that im working all the time with images that have an intuitive orientation (im ignoring the fact the officials are rotated, this simplifyes it and helps me to understand what im looking at), and im using as reference the image i made

I only worry about the rotations at the end, in the way i explained is automated with a bat, so no problem right now, and if at some point this feature to rotate jpeg images is added to the tool pink1 is making is going to be even more simple

--------------
Btw, not sure if some of you mentioned that link before, i was lurking it yesterday and they have photos of the earth clouds too (i guess better than the officials, not sure)
https://visibleearth.nasa.gov/collection/1484/blue-marble?page=1
The official specular images doesnt needs to be updated because the shapes matches perfectly with this new updated versions of the ground images (the official images probably are from 2006 but should match with the actuals from NASA unless we lost some island somewhere)

So... we can "update" the official images using this method, have you tryed it ?

*Btw, the images i used from ground have the "bathimetry"... are the areas under seas, cool effect (the officials doesnt have that, the areas with water sucks because looks like plastic)

 

[DeViL303]

Take another look at my previous post @DeViL303, i updated it to add the images generated after the process
About what we was discussing... one of the most important things of that process i explained is that im working all the time with images that have an intuitive orientation (im ignoring the fact the officials are rotated, this simplifyes it and helps me to understand what im looking at), and im using as reference the image i made

I only worry about the rotations at the end, in the way i explained is automated with a bat, so no problem right now, and if at some point this feature to rotate jpeg images is added to the tool pink1 is making is going to be even more simple

--------------
Btw, not sure if some of you mentioned that link before, i was lurking it yesterday and they have photos of the earth clouds too (i guess better than the officials, not sure)
https://visibleearth.nasa.gov/collection/1484/blue-marble?page=1
The official specular images doesnt needs to be updated because the shapes matches perfectly with this new updated versions of the ground images (the official images probably are from 2006 but should match with the actuals from NASA unless we lost some island somewhere)

So... we can "update" the official images using this method, have you tryed it ?

*Btw, the images i used from ground have the "bathimetry"... are the areas under seas, cool effect (the officials doesnt have that, the areas with water sucks because looks like plastic)

I would upgrade it if there was space. I am hoping pink1s tool will allow us to have like 10MB for this file.

I also am working with images the correct way up. I only rotate when I am injecting it. no point working with upside down images.

About what we was discussing... one of the most important things of that process i explained is that im working all the time with images that have an intuitive orientation (im ignoring the fact the officials are rotated, this simplifyes it and helps me to understand what im looking at), and im using as reference the image i made

I only worry about the rotations at the end, in the way i explained is automated with a bat, so no problem right now, and if at some point this feaure to rotate jpeg images is added to the tool pink1 is making is going to be even more simple

The rotation has never been an issue for me. its one click in an image editor or with a bat. I am not sure what is so important about the rotation or why even mention it?

My steps are

• take a cubemap or tile iamge
• rotate it,
• split it
• inject it.

I'm sorry, I do not really understand your last few posts at all.

This site makes perfect cubemaps at 4096x3072 and seem to be all I need for now.. https://matheowis.github.io/HDRI-to-CubeMap/ , it does not have any issues with the centre of images being in wrong place etc. All that is just more over complicating IMO...

 

[sandungas]

The "safe" texture resolution update to dont break whatever maths are made by openGL would be to increase the pixels along the equator line from the official 4096 up to 8192
In other words, the images with double the size, but thats a lot of difference, is not just some patches in the qrc structure, for that is going to be needed to rebuild it entirelly

Btw, something i didnt mentioned before... the qrc files are like an small variation of the CXML containers, the tool pink1 is doing is going to work with .qrc initially but eventually could work with .raf and .p3t too
The difference is in .raf and .p3t there is some special data (not files) stored inside
Processing that data is tricky, there are tables specific to store float values float arrays, etc...
And you need to extract it in some convenient format, usually .xml because this allows you to use that same data for rebuilding purposes
For .qrc you dont need that features... the only thing you need pink1 is to process the contents of the "file table" (and the id table to read the path/name.ext of the files)

That should be the priority... first deal with the file table (to suport full rebuilding of .qrc), and later eventually the other stuff if you want to continue working in it (.raf .p3t)

 

[sandungas]

I'm sorry, I do not really understand your last few posts at all.

Same process, except the fact we are using different reference images, this is what we was discussing wich orientation in the reference images could be more handy
My reference image have the same orientation than the "source" images except at the end
And im not using any custom name, after the cube faces are generated i label them with the official id, from 0 to 5, then cropped and renamed with a suffix with the bat to the official names

 

[DeViL303]

Ok, well like i said, having files with the same name is not convenient for me. So I use my own names. Its no different. just much simpler than 44 folder and 4x18 files with the same name.