Celestia Forums

Maybe I'm a little stupid, but I haven't been able to find a jovian DDS texture and the 4K jpeg and png-textures keep giving me headaches (OS X OpenGL bug)
Someone out there with a link?

Thanks in advance.

julesstoop wrote:Maybe I'm a little stupid, but I haven't been able to find a jovian DDS texture and the 4K jpeg and png-textures keep giving me headaches (OS X OpenGL bug)
Someone out there with a link?

Thanks in advance.

You are not stupid at all! I've been looking for the same thing a couple of months now and nothin'! I will also be grateful if someone pointed out on a link or something...

Hi,

What about processing your own dds from your 4k jpg or png??

I'm also looking for a jovian 8k, 16k or 32k map but seems to be some sort of rare piece...

(on osX graphic converter handle the dds format)

bye

I just uploaded a DDS VT with 2x2k tiles from the 4k Jupiter for folks who need 2k images for their graphics card. Hopefully it will be posted shortly.

See http://www.celestiamotherlode.net/creators/kolano/

Kolano wrote:See http://www.celestiamotherlode.net/creators/kolano/

Come on, that texture is known since years and is natively only 2k. Someone blew it up to 4k such that newbies feel they got the latest & greatest ...

More precisely, I think its original width was 2700 pix. So the serious guys (including myself about 2 years ago) reduced it to 2048 pix =2k and the 'extroverted' ones blew it up to 4096 =4k

Bye Fridger

Fridger,

Do you think such a map exist (8k or more) or there is not enough raw data (or something else) to make it?

ElChristou wrote:Fridger,

Do you think such a map exist (8k or more) or there is not enough raw data (or something else) to make it?

I said above: If anything /effectively/ bigger than 2k would exist for Jupiter, I (and others) would have made a corresponding Celestia texture since a long time.

Bye Fridger

AH was a test Fridger!!!
Don't you have better to do than responding such post (mine)?????

cf:---> something on cylindrical textures for exemple.

bye

ElChristou wrote:AH was a test Fridger!!!
Don't you have better to do than responding such post (mine)?????

cf:---> something on cylindrical textures for exemple.

bye

Since you keep pushing me: did you finish your homework?? Did you install GIMP and ImageMagick and test it on your OSX machine? Did you study the tutorial stuff that I posted as introduction?

You did not yet tell whether you know how to compile command line utilities under Unix/Linux...You did not yet tell what compiler you have.

Altogether, I guess there is quite a bit of work ahead of YOU before we can start ...


Bye Fridger

You know, since your last mail I'm on it, but it's not so simple for a neophyte like me (too much time with all done commercial programs...), so I don't sleep anymore, don't talk anymore, eat just the necessary and I said "adios" to outer world...
I don't know how long my wife will stay with me...

ok prof, magick is in, Gimp has been installed (It's really a photoshop like...)
Now the compile part... it's another story... well still diging...

Sorry perhaps I lose something... with tutorial???

Well your right, I'm far from the expert you are or better

bye

t00fri, thanks for the info. I had noticed there was little difference between my image and the 2k one. At the same time I use the 64k earth texture even though its based on only a 43k image, as I prefer to get all the detail I can, even if it means there are some wasted pixels. Though I guess I really should hit up the Nasa Worldwindow server at some point to grab the hi-res tiles there.

Kolano wrote:t00fri, thanks for the info. I had noticed there was little difference between my image and the 2k one. At the same time I use the 64k earth texture even though its based on only a 43k image, as I prefer to get all the detail I can, even if it means there are some wasted pixels. Though I guess I really should hit up the Nasa Worldwindow server at some point to grab the hi-res tiles there.

As should be obvious: interpolation is /much/ more reliable than extrapolation, notably if (bi) cubic interpolations are employed for the resizing process. To make a 4k texture of an original width of 2700 pixel, an /extrapolation/ by a factor 4096/2700=1.52 is required. In order to arrive at a 2k texture the rescaling /interpolation/ involved amounts only to 25%. (factor 0.75=2048/2700).

In case of earth, 32k <-- 43k --> 64k the factors are similar.

According to my personal experience with making hires 32k earth textures, many people are not aware how to sharpen correctly. The main problem in this game is NOISE as usual.

Incorrect sharpening tends to affect the colors and often makes things worse rather than better. It is crucial to decompose the colored earth texture into L A B grayscale channel layers first and only sharpen the luminance (L) channel!. Therafter one re-composes the L A B channels into the new RGB texture. This way, only the desired "signal" pixels are sharpened but NOT the noise! There are further tricks one may apply in addition, like making a smooth edge detection sharpening mask with standard image manipulation techniques. This way one may particularly and smartly sharpen sensitive areas in the texture. ...and so on...

Also, the sharpening must always be the /last/ operation.

In Celestia, the impression of the display most sensitively reflects the quality of the used normal map (much more than the main texture!). Thus, the main decisions on whether the result is pleasing or at best mediocre, implicitly take place when people perform image manipulation steps on the normal maps (16bit -> 8bit reduction, size reductions, sharpening, etc). An instructive selection of bad examples may be found on ML!

In summary:I claim that a well-done 32k earth texture, with all those points carefully accounted for, can compete well with a blown-up 64k appearance. Yet, let me emphasize, that Jestr's 64k earth is very nice!

Bye Fridger

t00fri wrote:Incorrect sharpening tends to affect the colors and often makes things worse rather than better. It is crucial to decompose the colored earth texture into L A B grayscale channel layers first and only sharpen the luminance (L) channel!. Therafter one re-composes the L A B channels into the new RGB texture. This way, only the desired "signal" pixels are sharpened but NOT the noise! There are further tricks one may apply in addition, like making a smooth edge detection sharpening mask with standard image manipulation techniques. This way one may particularly and smartly sharpen sensitive areas in the texture. ...and so on...

and so on... and so on what????

t00fri wrote:
Incorrect sharpening tends to affect the colors and often makes things worse rather than better. It is crucial to decompose the colored earth texture into L A B grayscale channel layers first and only sharpen the luminance (L) channel!. Therafter one re-composes the L A B channels into the new RGB texture. This way, only the desired "signal" pixels are sharpened but NOT the noise! There are further tricks one may apply in addition, like making a smooth edge detection sharpening mask with standard image manipulation techniques. This way one may particularly and smartly sharpen sensitive areas in the texture. ...and so on...



Bye Fridger

t00fri , regarding the sharpening of only the L - channel: doesn't all of these advanced image manipulation programs
like Paint Shop or PhotoShop do all this for you if you just apply sharpening ??


RRA

I will respond you cause Fridger is quit busy; nope, those programs apply the filter on all the RVB layers... and using the LAB mode make a great difference, you have to do the test with a piece of a reduced map (16k > 4K)...

ElChristou wrote:I will respond you cause Fridger is quit busy; nope, those programs apply the filter on all the RVB layers... and using the LAB mode make a great difference, you have to do the test with a piece of a reduced map (16k > 4K)...

Thanks ElChristou,

you answered exactly what I would have written

Bye Fridger

ElChristou wrote:I will respond you cause Fridger is quit busy; nope, those programs apply the filter on all the RVB layers... and using the LAB mode make a great difference, you have to do the test with a piece of a reduced map (16k > 4K)...

ElChristou , I will try it ,
I noticed that PSP doesn't support splitting into LAB planes,
I presume that splitting into HSL and sharpenig the L plane out of HSL
would yield the same result .

Thanks ,

RRA

You can test it but don't forget PSP is not a "profesional" program, just a paint ++... Better use The Gimp (free) if you cannot use photoshop.

Bye

ElChristou wrote:Fridger,

Do you think such a map exist (8k or more) or there is not enough raw data (or something else) to make it?

The biggest referenced jupiter map is about 5k. Bj?¶rn J??nsson claims to have made one:

This map of Jupiter was created from roughly 90 images taken by the Voyager 2 spacecraft at the end of June 1979 from a distance of a little under 9 million km. These images have a resolution of roughly 90 km/pixel so to match this the original map has a size of 5040x2520 pixels (14 pixels per degree) but the version here is 1800x900 pixels.

AFAIK it was never published.

maxim