Celestia Forums

As suggested here:

http://celestiaproject.net/forum/viewtopic.php?p=92810#92810

this thread concern the creation of texture packs for Celestia to be hosted on Motherload in substitution of thats shipped with the main distribution.

To summarize, I can test only the 2k, so the rest is yours. Once collected, will be assembled the relevant resolution packs.

DIRECTIVES...

1) The opening goal for all the Celestians is to link here the better textures available for the aforementioned resolutions, early TESTED when dowloaded by the user, alphabetically sorted and relate exclusively to that previous shipped (i. e. no so many textures for so many asteroids, thus the same for exoplanets and NO FICTIONALS);

2) The textures may be: surfaces, normals, speculars, clouds, nights, lokmask, rings, stars (no bumpmaps, please; double GPU calculus) both .PNG and DDS (MacOS and Linux users .DDS are optional) no VT, for the moment; the textures name may be lower case for all.

3) Who agree, please be sure of write of what body concern (i.e: amalthea, then the link at). Does can be mentioned the same textures, which will have the favourite pick out just in the case the choices to be greater of three; only one for rare textures (who has seen the 8k of ariel?)

4) What lack to all the points to be complete;

5) As start, the statements 1) - 2) - 3) - 4) are under discussion.

and i mite add NO .dds for us Linux and i think also mac users

-- just a rant --
also a common spelling / capitalization for us that use a os that is
CAP sensitive

ie. EarthNornal.png and NOT earthnornal.PNG or EARTHNORMAL.PNG
i hate having to run my case script and figuring out witch way it is spelled in the cmod or 3ds .... more ...
----end of rant ---

john Van Vliet wrote:and i mite add NO .dds for us Linux and i think also mac users

-- just a rant --
also a common spelling / capitalization for us that use a os that is
CAP sensitive

ie. EarthNornal.png and NOT earthnornal.PNG or EARTHNORMAL.PNG
i hate having to run my case script and figuring out witch way it is spelled in the cmod or 3ds .... more ...
----end of rant ---

Accepted and corrected the directives. Lower case for all.

dds is fine for mac... well... they work for me.

bh wrote:dds is fine for mac... well... they work for me.

Well, I correct as "optional".

What sign? E.a:
amalthea_*;
amalthea-*;
*_normal.*;
*-normal.*;
*_normalmap.*;
*-normalmap.*;

the rest it would be identical, I believe:
- or _ and specular.*; night.*; clouds.*; rings.*;lok(_ or -)mask.*


I've forgot something?

Fenerit,

please, don't kill me, but I think this job is just too big for you.

What do you think, why after 6 years of Celestia we have not yet arrived at such a texture pack? It was certainly NOT because noone did have that idea before...

Such a challenging task should be done and|or guided, by a consortium with proven expertise in image manipulation and Celestia texture making! Otherwise, it's really not worth the large effort. It should involve experts who can tell within a short time whether a texture is for the dustbin or not. It should involve people aware of the pros and cons of recent compression software developments and of course with knowledge about planetary science.

The various new developments in this important field of high-speed and high-quality, hardware supported texture compression should NOT be ignored, of course. The new OpenSource NVIDIA nvcompress, based on the high-quality DXT compression squish is available for all operating systems. There should be UNIFORM guidelines for the texture pack in this respect. The composition of a GOOD texture pack needs EXPERT knowledge about these new, exciting options!

Furthermore, there are NEW toolsets that will allow generation of textures in reproducable highest quality directly from scientific imaging data (nmtools, F-TexTools) .

Respective efforts are discussed at our Celestial Matters forum.

I neither have the time for participating in your project, nor do I give it much chances for real success.

Still, good luck for everyone involved in it,

Bye Fridger

Fridger, we are agree on this, but nothing cost trying. Also it's a good way for Fenerit (and all of us) to learn what implies such packs...

Now of course, if the result is not 100% satisfactory, it can stay at addon level, but will be welcome by newbies who don't know what to download in the Motherlode sea of files...

ElChristou wrote:Fridger, we are agree on this, but nothing cost trying. Also it's a good way for Fenerit (and all of us) to learn what implies such packs...

Trying in vain costs a lot of time and brings frustration...

Now of course, if the result is not 100% satisfactory, it can stay at addon level, but will be welcome by newbies who don't know what to download in the Motherlode sea of files...

As I said, good luck to anyone who wants to embark on this at a low know-how level. I just wanted to express, that the involved people can hardly learn much without the proper experts on board...

Limiting such a pack to 2k is not so bad, since Newbies don't see much difference at the 2k level anyhow


Bye Fridger

t00fri wrote:Fenerit,

please, don't kill me, but I think this job is just too big for you.

What do you think, why after 6 years of Celestia we have not yet arrived at such a texture pack? It was certainly NOT because noone did have that idea before...

Such a challenging task should be done and|or guided, by a consortium with proven expertise in image manipulation and Celestia texture making! Otherwise, it's really not worth the large effort. It should involve experts who can tell within a short time whether a texture is for the dustbin or not. It should involve people aware of the pros and cons of recent compression software developments and of course with knowledge about planetary science.

The various new developments in this important field of high-speed and high-quality, hardware supported texture compression should NOT be ignored, of course. The new OpenSource NVIDIA nvcompress, based on the high-quality DXT compression squish is available for all operating systems. There should be UNIFORM guidelines for the texture pack in this respect. The composition of a GOOD texture pack needs EXPERT knowledge about these new, exciting options!

Furthermore, there are NEW toolsets that will allow generation of textures in reproducable highest quality directly from scientific imaging data (nmtools, F-TexTools) .

Respective efforts are discussed at our Celestial Matters forum.

I neither have the time for participating in your project, nor do I give it much chances for real success.

Still, good luck for everyone involved in it,

Bye Fridger

Indeed. I should to end my add-ons in progress, so I've put down on the road just the directives, because I think it can be follow from everyone who like to assemble a pack for own count, and at the same time I can't stay indefinitely in front of a computer. I've made the 2k and I'd the intention of maintain it from time to time until all the basic textures would have to match as the same of the official, basic ones (plus normalmaps) and in doing this usually I use the 4k or 8k resized by myself, by enlarging > deconvoluting for that which are unavailables (1k Uranus is the maximum on Motherload!) or waiting for the news enhanced, and I'm happy if someone attempt to test it, how has done DanielJ. So thanks Fridger for these remarks; it help me to buoyed up at my pertinence level.

Fenerit wrote:I use the 4k or 8k resized by myself, by enlarging > deconvoluting for that which are unavailables (1k Uranus is the maximum on Motherload!)

Fenerit,

as far as I am aware, there is NO scientific Uranus texture bigger than 1k. Were you saying that you just resize the 1k texture to 4k or even 8k???

This must be a misunderstanding, right?

Bye Fridger

t00fri wrote:

Fenerit wrote:I use the 4k or 8k resized by myself, by enlarging > deconvoluting for that which are unavailables (1k Uranus is the maximum on Motherload!)

Fenerit,

as far as I am aware, there is NO scientific Uranus texture bigger than 1k. Were you saying that you just resize the 1k texture to 4k or even 8k???

This must be a misunderstanding, right?

Bye Fridger

I've undestood what you mean about the meaning of "scientific", but on other hand there is not misunderstanding. My "enlargement" is just empirical, for visual purposes. I split the 1k in 4 parts, then I use each part as procedural texture in 3d modeller by mapping a plane; this operation raise the resolution, then reassemble the pieces at 8k and finally resize all at 2k at 1200dpi. If at closest approach 1k show the grains, in this way at same distance i do not see the grains.

Fenerit wrote:

t00fri wrote:

Fenerit wrote:I use the 4k or 8k resized by myself, by enlarging > deconvoluting for that which are unavailables (1k Uranus is the maximum on Motherload!)

Fenerit,

as far as I am aware, there is NO scientific Uranus texture bigger than 1k. Were you saying that you just resize the 1k texture to 4k or even 8k???

This must be a misunderstanding, right?

Bye Fridger

I've undestood what you mean about the meaning of "scientific", but on other hand there is not misunderstanding. My "enlargement" is just empirical, for visual purposes. I split the 1k in 4 parts, then I use each part as procedural texture in 3d modeller by mapping a plane; this operation raise the resolution, then reassemble the pieces at 8k and finally resize all at 2k at 1200dpi. If at closest approach 1k show the grains, in this way at same distance i do not see the grains.

Fenerit,

but calling such an approach a 4k or 8k image is highly misleading.

There are soooooooooooooo many powerful (professional) filter programs out there that get rid of grains without increasing the filesize by a factor 16 or even 64 (!!) compared to the original image

And I am sure you know that all that counts is the amount of detail which cannot be increased in any way by just blowing up the size of an image!

Bye Fridger

Yes. For what concern the details there is nothing to do scientifically, of course. Mine are little visual improvements of the extants. Sincerly, I do not know better way to do it, nor what programs does as you say. Do you for courtesy tell me some names? I suspect no one to be freeware!

Another one about texture packs. I believe there is even a "fairplay" question about the lack of decision in creation of; that is, the will of no prefer one author to another and the modesty of authors themselves, who are reluctants in pick out the proper work.

But Fenerit, is there really an improvement between your 1k and your 8K (extended from the 1k)? Is the improvement worst the extra resource use?

ElChristou wrote:But Fenerit, is there really an improvement between your 1k and your 8K (extended from the 1k)? Is the improvement worst the extra resource use?

Worst no. I'll try to explain better what I intend. If I enlarge the 1k to 8k and then resize to 2k, (the maximum for my graphic card) at maximum dpi, to say 1200, doesn't happen nothing, as you well know; but with a slighty blur on 8k when risized to 2k is little better (in visual only, not "scientific" as I've cleared). But the metod of procedural texture is better again even with the entire texture; in splitting it, is better more. You can try: take a 1k texture and apply it on a 2 x 1 plane, then tell at your modeller to create an image output of 8k. Then does import this texture in a phoretouch program and resize it at 1k with the maximum dpi. Try now to zoom in both the 1k textures; the old and the new. At 200% in the old you'll see the chess (I do not have the right word for this now, sorry) in the new no; you'll see it around the 600%.
Again, when I make a 1024x1024 pixel gradient in Photoshop at 300dpi for my Mantle cross-section, when I am at closest approach I see the grains, while I do not see they when I make the same texture in procedural way of 256x256 pixels.

EDIT LATER:
You can see that (for the Mantle) in observing the gray bend cube of the 220 discontinuity. That is a procedural texture. You'll do not see the grains.

Unfortunately this computer is not that on which I work, then I'll post the Uranus image for both procediments take at same distance.

Fenerit,
i think your project is commendable.
However, if you limit the texture collection to limit-of-knowledge, it will be much more valuable even if it might not look as good. You can let it be up to the indivdual user to add eyecandy modifications.

t00fri wrote:

ElChristou wrote:Fridger, we are agree on this, but nothing cost trying. Also it's a good way for Fenerit (and all of us) to learn what implies such packs...

Trying in vain costs a lot of time and brings frustration...

It may bring frustration, but also experience...

- rthorvald

PS: i saw your website. Thank you for the link!

rthorvald wrote:Fenerit,
i think your project is commendable.
However, if you limit the texture collection to limit-of-knowledge, it will be much more valuable even if it might not look as good. You can let it be up to the indivdual user to add eyecandy modifications.

t00fri wrote:

ElChristou wrote:Fridger, we are agree on this, but nothing cost trying. Also it's a good way for Fenerit (and all of us) to learn what implies such packs...

Trying in vain costs a lot of time and brings frustration...

It may bring frustration, but also experience...

- rthorvald

PS: i saw your website. Thank you for the link!

Yes, indeed. Mind that my operation are to raise from 1k to 2K only; from 8k or 4k to 2k I use the best yours (i.e. your Saturn), by resizing at maxim dpi for fidelity when I do not find the 2k already extants.

Just to add more: when the Nvidia DDS plugin converter act on 2k .PNG at 1200 dpi, it spend ten time in process with respect of a 2k at 72 dpi.

Fenerit wrote:from 1k to 2K only; from 8k or 4k to 2k I use the best yours (i.e. your Saturn), by resizing at maxim dpi for fidelity

While you certainly are welcome to use my Saturn in any way you like, it is a case in point: my Saturn is not limit-of-knowledge. It is an interpretation, where i used Ciclops imagery to enhance Bjorn??s original Voyager map. Thus, while it is a more pleasing and detailed map of Saturn than the other ones available, the source is the more exact one... So that is what you should use if you want to produce a scientifically valid package.

Another candidate is of course Fridgers 2k "blue hat" from 2005:
http://celestiaproject.net/forum/viewtopic.php ... ght=saturn

- Runar