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

[sandungas]

Honeycomb earth.qrc!!

Thats what i mean with the term "tiles" btw, a single small image of any size that should be spreaded along the whole sphere with the same size
In this case the tile is an hexagon... but we cant display all instances of it at the same size :/

In your image none of the hexagons have a perfect geometry... the ones in the equator are "expanded" and the ones in the poles are "squeezed"... so the only ones that gets closer to be a perfect hexagon are the ones along the tropic lines (but not even because his distortion follows the same rules than all the others... is just is not much notable in that ones)

 

[xf505]

Alien Artifacts earth.qrc!! This will be my last one for awhile. Goodluck everyone and have fun creating them, I know I did!! Thanks @DeViL303 for making it EASY!!

 

[xf505]

Thats what i mean with the term "tiles" btw, a single small image of any size that should be spreaded along the whole sphere with the same size
In this case the tile is an hexagon... but we cant display all instances of it at the same size :/

In your image none of the hexagons have a perfect geometry... the ones in the equator are "expanded" and the ones in the poles are "squeezed"... so the only ones that gets closer to be a perfect hexagon are the ones along the tropic lines (but not even because his distortion follows the same rules than all the others... is just is not much notable in that ones)

I personally like the imperfections so I guess it's all up to the creator and I'm not spending more time on these than I already have so Goodluck getting them "perfect"

 

[sandungas]

I mean... we cant do things like this, im not critizising what you are doing... im just showing the problem to talk about it
Creating things like this is a geometry challengue that i like, this is why i been so interested in the distortions

I dont like soccer, and im not going to make a mod like that... im just using that image because is a great example of something we cant do accuratelly

 

[pink1]

After I was in bed last night it hit me I didn't write the new file table size back into the header
So far it's looking better now lets hope the rest of the test go good.

 

[DeViL303]

After I was in bed last night it hit me I didn't write the new file table size back into the header
So far it's looking better now lets hope the rest of the test go good.

YEAH, its perfect man. Injected some random 127kb jpgs into about 6 slots. Now running an oversized 2.22MB earth.qrc with high bitrate images.

You the man, this changes everything now.

 

[DeViL303]

The images already look so much better , wow.

I bet the guys at q-games were so disapointed to have to have limit the quality so much in the final release.


see this interview, they were worried people would not like it as they had to compress it so much...

https://www.gamasutra.com/view/news/107699/Special_QGames_On_PS3s_Gaia_Music_Visualizer.php

We are basically making it the quality it was MEANT to be originally.

 

[pink1]

That's pretty funny. I'm interested to see what we can do for the other CXML formats.

I'm honestly so happy that it actually works lol

 

[DeViL303]

That's pretty funny. I'm interested to see what we can do for the other CXML formats.

I'm honestly so happy that it actually works lol

Next feature request Can you make an easy way to dump in a complete cubemap to each of the 3 layers?

 

[DeViL303]

Even if i could just extract all 72 jpegs fully to the official file structure, and then rebuild from those jpgs.

This way I can use the output from my bat by copy and pasting them in.

 

[sandungas]

Even if i could just extract all 72 jpegs fully to the official file structure, and then rebuild from those jpgs.

This way I can use the output from my bat by copy and pasting them in.

Maybe by adding a column of checkboxes at the left of the file list, then "extract selected files"

Next feature request Can you make an easy way to dump in a complete cubemap to each of the 3 layers?

But thats jpegtran task... we was talking if it could be posible to integrate it in the qrc editor, but you know... by now "we dont know" how to do it
That image is usually named "cubecross" btw, the tool from ATI/AMD uses that name and i found other places using it too, his orientation uses to be like you do, but ive seen others where is used rotated in different ways, as example it seems in the tool from ATI/AMD are vertical like a chatolic cross

 

[pink1]

Next feature request Can you make an easy way to dump in a complete cubemap to each of the 3 layers?

I'm sure we can get something going for that.
I think today I'll try to get some of the image stuff added and put off creating a qrc from a folder for now.

 

[DeViL303]

Maybe by adding a column of checkboxes at the left of the file list, then "extract selected files"

Yeah, but too slow if i need to click 72 boxes, I need EASY.

But thats jpegtrans task
That image is usually named "cubecross" btw, the tool from ATI/AMD uses that name and i found other places using it too, his orientation uses to be like you do, but ive seen others where is used rotated in different ways, as example it seems in the tool from ATI/AMD are vertical like a chatolic cross

Yes that is for jpegrans, but I think you still do not know what is involved in this, PLEASE try adding 72 images now using the QRC editor and see how long it takes you. please do it, i will give you 72 preprepared images if you want. It is amazing already, but is not convenient. simple as that.

 

[sandungas]

Yeah, but too slow if i need to click 72 boxes, I need EASY.

Then you need other option named "select by filetype"
Doing it this way is more generic (not only for earth), it will work with all other .qrc files

Yes that is for jpegrans, but I think you still do not know what is involved in this, PLEASE try adding 72 images now using the QRC editor and see how long it takes you. please do it, i will give you 72 preprepared images if you want. It is amazing already, but is not convenient. simple as that.

The program CubeTheSphere generates the cube faces already cropped in 6 images, thats a convenient way to do it too (and it doesnt involves using the intermediate "cubecross" format)

And the orientation of that cubecross is not an standard... every developer/program does whatever they wants, see how is made in skyrim with the cross in vertical
https://wiki.nexusmods.com/index.php/Creating_cubemaps_for_Skyrim

...and what sony did in the PS3 earth is not standard, they decided to rotate the cubecross 180

 

[pink1]

Yeah, but too slow if i need to click 72 boxes, I need EASY.


Yes that is for jpegrans, but I think you still do not know what is involved in this, PLEASE try adding 72 images now using the QRC editor and see how long it takes you. please do it, i will give you 72 preprepared images if you want. It is amazing already, but is not convenient. simple as that.

I was thinking about maybe having three buttons, one for each layer. For replacing the images I was thinking about having a xml or something that would have a list of the images you're replacing and the file name of your new images. I thought doing it that way we can get it the way we like and everyone else can change it if they want.
I'm open to ideas that's just some thoughts I had.

 

[DeViL303]

Yeah, I think we should work with "normal" cubemaps. 95% of cuemaps and skyboxes available to people are in this format, I do not care about sony using upside down ones it makes no difference to me.. Do a google search for cubemap see how many upside down ones you find. The simple thing IMO is to use the normal map and rotate it before injection

The way I made my bat is to ALWAYS make a cubemap.. even when being given face.jpg or tile.jpg....Some might say this is not optimized and is a waste of time....but in real use it is super useful to be able to refer to the cubemap and see where each image you gave it is being placed.

For that reason I think any images given to the editor should be converted to cubemap first at 4096x3072..then split back down into each jpeg.

 

[DeViL303]

I was thinking about maybe having three buttons, one for each layer. For replacing the images I was thinking about having a xml or something that would have a list of the images you're replacing and the file name of your new images. I thought doing it that way we can get it the way we like and everyone else can change it if they want.
I'm open to ideas that's just some thoughts I had.

To be honest now that i have my bat. I can work with the official file structure without even having to see it. The best thing I think would be to be able to rebuild from a folder structure. That allows complete customization.

For actual image editing then I think work with cubemaps... if anything smaller is given to the app, create a cubemap and keep that as a preview for the app. Then split it back down.. they are so valuable when you make a mistake and place an image wrong.

Also you can copy paste all of that from my bat if you want to do it with jpegtrans

 

[sandungas]

Whatever you do @pink1 i think you should not drop the support to import images in "cube face" format

The point is with a program like cubethesphere we are converting images from the format used in NASA maps to cube faces directly, so is like this:
Spherical map ---> 6 * cube faces
And in the web services is like this:
Spherical map ---> cubecross ---> 6 * cube faces

I think thats a really good argument, but also consider that web services needs online and doesnt works in old pc's with old web browsers

 

[DeViL303]

True, I think it should be able to guess from the width to height ratio and the resolution what is it s dealing with and should be able to create a cubemap out of anything that is not a cubemap already. My bat can already do that, but it is not "smart" , it just does that based on fileaname.

Also my bat is missing that one important step, being able to handle flat maps.

Maybe it should work something like this, have folder to qrc creation for advanced users. But also have it set up to try be smart for noobs.

So maybe if you give it a 4:3 ratio image that is like 4096x3072 etc assume cubemap. If you give it a 2:1 ratio image that is like 4096x2048 , 8192x4096 etc then assume its a flat map that needs to be converted to cubemap, either warn user its probably going to look crap or maybe do the conversion.

If you give it a 512x512 image assume its a tile. if you give it a 1024x1024 assume its a face. if you give it a 1:1 ratio width/height image that is NOT 512x512 then assume its a face.

All these assumptions are only for default handling to help noobs, you could add the ability to over ride them and apply any image in any way. For example force a 512x512 image to be stretched over the whole globe like a flat map.

 

[DeViL303]

Successfully running a 5.64MB earth.qrc now Complete cloud layer replaced with 215KB images.

I will brute force the max size by trial and error. Unless there is none