Ceres Textures

nardo · Post #21by **nardo** » 21.10.2016, 17:37

FarGetaNik wrote:That's possible on USGS?

map2 on usgs gives you some basic tools to get the image you want in the format you want. it took about a day for my account to get approved. here, i'll upload some screenshots of the main tools.

FarGetaNik · Post #22by **FarGetaNik** » 21.10.2016, 18:23

Thanks for the tip, I just requested an account :smile:

let's see if this works...

John Van Vliet · Post #23by **John Van Vliet** » 21.10.2016, 21:03

the Dev gimp 2.9 supports 16 and 32 bit images
the linux build is newer but the windows one is supposed to work well

the images i posted a few posts back are using Nip2 ( runs on Linux, Mac, and Win 8 & 10 )
and nip handles 16 and 32 bit images and is DESIGNED for use with HUGE image sets

i have no problems with a 24 gig 32 bit height map
and i only have 8 gig ram

PS i used nip2 to fix the low light at the poles

nardo · Post #24by **nardo** » 21.10.2016, 22:02

would you know if either could take a function like a contrast/brightness change and apply it as a linear or radial gradient?

John Van Vliet · Post #25by **John Van Vliet** » 21.10.2016, 22:15

that is what i did with the tile lighting - part of the mosaic tools for relighting an array of tie point blended images
( or one image)
-- Nip2

i just made a 1024x1024 pixel image with a value of 2 and used a circular light

and gimp with the gmic plugin
a 32 bit floating point radial gradient in gimp 2.9.4

nardo · Post #26by **nardo** » 21.10.2016, 22:41

John Van Vliet wrote:that is what i did with the tile lighting

awesome! that's what i had in mind for some of the junocam images as well.

FarGetaNik · Post #27by **FarGetaNik** » 22.10.2016, 07:48

John Van Vliet wrote:the Dev gimp 2.9 supports 16 and 32 bit images
the linux build is newer but the windows one is supposed to work well

Well more or less. My system cosnstantly confuses gimp 2.8 and gimp 2.9, which results in long startup times. And sometimes (quite regularly with these huge textures infact) a glitch occours and destroys the image... some artifacts/noise is created and won't disappear with edit/undo.

I guess I'll try Nip2 then.

Added after 18 minutes 11 seconds:
Could you provide a download link? I can't find a working, trustworthy Windows installation online. Only source code or broken download mirrors.

nardo · Post #28by **nardo** » 22.10.2016, 11:34

i used this

FarGetaNik · Post #29by **FarGetaNik** » 22.10.2016, 11:42

I guess I have to trust this, my "Web of Trust" plugin is suspicious of this site and indicated potential malware :spy:

that's why I was ignoring this site at first.

nardo · Post #30by **nardo** » 22.10.2016, 12:38

hmm. well if it gives you problems i could always just give you the setup file, but it should be fine.

John Van Vliet · Post #31by **John Van Vliet** » 22.10.2016, 18:09

why are you using a rather OLD version
and from a KNOWN malicious web site

that version 722 is from 2014

http://www.vips.ecs.soton.ac.uk/index.php?title=VIPS

the current windows version is 8.4
http://www.vips.ecs.soton.ac.uk/supported/current/win32/vips-dev-w64-all-8.4.2.zip

check your windows registry for NEW entries for the NEW virus/keyloger/botnet that was installed
and run mcaffe's online scan

Added after 20 minutes 3 seconds:
and here is a link to a very good blog on nip/vips
http://libvips.blogspot.com/

and the documentation
http://www.vips.ecs.soton.ac.uk/index.php?title=Documentation

nardo · Post #32by **nardo** » 22.10.2016, 18:29

John Van Vliet wrote:why are you using a rather OLD version
and from a KNOWN malicious web site

because all the download links for binaries are broken.

John Van Vliet wrote:the current windows version is 8.4

thanks! i stand corrected.

John Van Vliet wrote:check your windows registry for NEW entries for the NEW virus/keyloger/botnet that was installed
and run mcaffe's online scan

this is a new laptop so i keep it really clean don't worry

edit: ok now i'm confused. where is nip2. this might end up needing its own thread.

John Van Vliet · Post #33by **John Van Vliet** » 22.10.2016, 19:14

apparently there is a issue with 8.4
use 8.3
http://www.vips.ecs.soton.ac.uk/supported/8.3/win32/
-- the gui nip
http://www.vips.ecs.soton.ac.uk/supported/8.3/win32/nip2-8.2b-setup.zip
and the lib vips
http://www.vips.ecs.soton.ac.uk/supported/8.3/win32/vips-dev-w64-web-8.3.3a.zip

things windows might need if it complains are also in that folder
http://www.vips.ecs.soton.ac.uk/supported/8.3/win32/

and a giude
http://www.vips.ecs.soton.ac.uk/supported/8.3/doc/html/nipguide/nipguide.html

FarGetaNik · Post #34by **FarGetaNik** » 23.10.2016, 08:15

Ok this all just added to the confusion, but I was able to install and run nip2 8.2. Now I will have to find out how to use it as an image processing program... this might take a while.

FarGetaNik · Post #35by **FarGetaNik** » 31.05.2017, 13:49

So I found the problem with the USGS maps (like this DEM of Ceres): The tiff files for download are in 16 bit signed integer format. Neither Gimp nor Photoshop can handle this apparently, but NIP2 does! Luckily, when exporting these textures to NIP2's internal .vips format, I was able to create a proper normalmap with t00fry's texture tools. I think I created a normalmap of realistic intensity if I used the tools correctly. It creates some slight artifacts though (a 31 pixel strip is shifted by 1 pixel) but I was able to correct that. Also I converted the tiff file into a 16bit unsigned integer png file by a rather complicated process and removed the effects of Ceres' oblateness on both the bump and the normal map.

With the normalmap I was able to remove the shadows in this color map.

Note that all these textures are in planetocentric projection and thus must be rendered on a mesh for correct projection. This is what Ceres looks like now:

Bump-Map (16k png, 16bit color depth, 141 MB):
https://drive.google.com/open?id=0B14NSu0PnA5xMmxEc1d3d2trLVU

Normal-Map (16k png, 117 MB):
https://drive.google.com/open?id=0B14NSu0PnA5xTk9aM2hGc2dybjQ

Color Map (not calibrated):
https://drive.google.com/open?id=0B14NSu0PnA5xVnhPbHdiMmJJYm8

3D model I used (HAMO 256k poly):
https://space.frieger.com/asteroids/dwarf/1-Ceres

Also I was able to create a 64k normal map/virtual texture of the moon, as well as a normalmap for Mercury. Unfortunately, the Mars DEM seems to not cover all latitudes up to 90°(in contrast to what USGS claims), so I had to scrap it.

fsgregs · Post #36by **fsgregs** » 31.05.2017, 14:32

Great looking stuff. Are you planning (I hope) to create a mesh we can use in Celestia? I have no idea how to do so.

Frank

FarGetaNik · Post #37by **FarGetaNik** » 31.05.2017, 15:03

fsgregs wrote:Great looking stuff. Are you planning (I hope) to create a mesh we can use in Celestia? I have no idea how to do so.

I already did. Here you go

I also made a tutorial on how to put a model in Celestia: https://celestiaproject.net/forum/viewtopic.php?f=21&t=17427

fsgregs · Post #38by **fsgregs** » 31.05.2017, 16:04

thanks

Looks great

Frank

John Van Vliet · Post #39by **John Van Vliet** » 31.05.2017, 16:39

why do you think i have been using Nip for all these years
gmic also handles 16 bit signed integer images

FarGetaNik · Post #40by **FarGetaNik** » 31.05.2017, 18:27

Yes I get it. I still did most of the processing in Gimp though. NIP2 is somewhat unfamiliar to use, so I don't know how to do advanced tasks. Well I hope that a stable version of Gimp 2.10 on good hardware can do the job; I have no idea how to remove the shadows in NIP2 as I did in Gimp. I'd like to to that on a 23k texture (or 16k), but Gimp 2.9 can't hadle this on my hardware right now.

Come on, I finally managed to use t00fri's texture tools, took me long enough :zombie: