Printable Version of Topic

Posted by: elakdawalla Jan 29 2016, 03:58 PM

Chang'e 3 data is now available to the international community in PDS format through http://moon.bao.ac.cn/ceweb/datasrv/dmsce3.jsp. The website tends to be unstable; login has not worked for me all week...but it is working now. If you don't want to work through the website, http://www.planetary.org/blogs/emily-lakdawalla/2016/01281656-fun-with-a-new-data-set-change.html. Have at it!

Posted by: Phil Stooke Jan 29 2016, 05:54 PM

"Have at it!"

Oh, OK, if I must...


Emily posted a partial panorama with the lander in the blog post. Here is a cropped version of it with a vertical stretch to show horizon features better. In particular, it shows two distant hills poking up above the rim of the nearby crater Tai Wei. I identify them with a crater 15 km to the NNE. It is shown in the accompanying map (part of my new project) as well as other craters including 'northeast crater', also visible in the stretched panorama.

Phil

Posted by: jccwrt Jan 30 2016, 01:29 AM

I've taken a stab at working with this data set using the demosaiced png images. Colorization was done using gmic's "bayer2rgb" routine, which for some reason spit these images out with the red channel significantly fainter than the green and blue channels. As a result I've kind of had to guess at how bright the red needs to be, so the color is probably off. I should probably grab some of the official releases to compare color.

Anyway I did some cleaning up in Photoshop (slight gaussian filter to remove the last traces of bayer patterning, a denoise filter to remove some of the color noise, and some sharpening to recover some of the detail lost with the bayer pattern removal) and here's the result:

https://flic.kr/p/Dbo71h
https://flic.kr/p/Dbo71h by https://www.flickr.com/photos/132160802@N06/, on Flickr

The problem I'm trying to tackle now is a slight color gradient between the left and right edges of the images. Although Photomerge does a good job handling overall brightness gradients, it's having a little more difficulty with the color. The neutral gray color of the terrain also makes slight variations in color stick out more. Here I've tried to reduce it with a bit of burning and dodging but I'm not entirely happy with the result. Once I get a better process down I'll probably rework this image, as well as assemble the full 360 degree panorama.

Posted by: elakdawalla Jan 30 2016, 01:34 AM

I've had the same problem with the TCAM data -- they come out very cyan. The PCAM data look much, much better. I was also getting the sense that the quality of the TCAM images was degrading over time on the lunar surface, but I didn't check that systematically.

I think that the gradient is a photometric effect. In the past I've tried to deal with that by opening the images in Photoshop and playing with levels in individual frames to make things roughly consistent across all images in a panorama, and *then* using Photomerge or ICE, rather than relying on the automated algorithms to do all the work of matching the color across the panorama.

Posted by: Hungry4info Jan 30 2016, 12:46 PM

I figured I would have a go at making some 3D anaglyphs.
(thank-you so very much for hosting these in an easily accessible site!)

Posted by: jccwrt Jan 31 2016, 01:49 AM

Attempt #2 with the Observation 004 TCAM images. To get a better adjustment on the color, I selected a few points from a Yutu image and adjusted the relative brightnesses of the color channels in the lander camera images to match. I have a slightly better handle on the photometric color variation across each subframe, although if you look closely you'll still see some uneven cyan/green blotching here and there.

https://flic.kr/p/DMd8Lh
https://flic.kr/p/DMd8Lh by https://www.flickr.com/photos/132160802@N06/, on Flickr

And in polar view:
https://flic.kr/p/DPxdTk
https://flic.kr/p/DPxdTk by https://www.flickr.com/photos/132160802@N06/, on Flickr


EDIT (Feb 4): replaced images with red/blue channels reversed

Posted by: jccwrt Feb 2 2016, 11:37 PM

Full panorama from Yutu Observation 008:

https://flic.kr/p/Dc9tYX
https://flic.kr/p/Dc9tYX by https://www.flickr.com/photos/132160802@N06/, on Flickr

Polar view:

https://flic.kr/p/DKD5Cp
https://flic.kr/p/DKD5Cp by https://www.flickr.com/photos/132160802@N06/, on Flickr

Posted by: elakdawalla Feb 2 2016, 11:46 PM

Beautiful!! Here it is on roundme: https://round.me/tour/27171/edit/65482/

Now make it stereo

Posted by: jccwrt Feb 4 2016, 01:07 AM

Ask and you shall receive!

https://flic.kr/p/DKsBzC
https://flic.kr/p/DKsBzC by https://www.flickr.com/photos/132160802@N06/, on Flickr

I'm having difficulty getting both the foreground and background to "focus" at the same time. Part of the problem is that I don't have a pair of 3D glasses, I've been working on it looking through light red and light blue astronomical filters to get an idea of what it looks like. Although I can start to see the 3D effect, there are a bunch of artifacts from incomplete blockage of the red and cyan images, so it's hard for me to tell what's just an artifact or what's coming from bad alignment.

Just moving the left eye images around by trial and error doesn't seem to produce an optimal result. The convergence point in the raw left-right pairs appears to be only slightly behind Dragon Rock, so trying to align the images on a feature in the foreground produces unacceptable amounts of parallax in the background, and vice versa. So I just aligned on the convergence point of the center pair in this three image mosaic.

Photoshop really isn't equipped to handle making anaglyphs from multiple images, so I probably need to find an alternative program to make them in the future.

Posted by: elakdawalla Feb 4 2016, 01:19 AM

I wonder if you'd have an easier time trying a different set of images? This is a particularly weird one, with a huge deep hole right behind a great big foreground rock.

Posted by: jccwrt Feb 4 2016, 01:50 AM

Yeah, I'm going to try and take another shot with some images with the lander in it - they're mostly flat and hopefully the parallax problems shouldn't be as severe. If not, I don't think I'll be able to make good anaglyphs with this data set.

Posted by: machi Feb 4 2016, 03:25 AM

Thanks Emily for great gallery of Chang'E 3 images. It's very helpful.
Beautiful mosaics Justin, but I have one point - evidently TCAM camera has different order of colors in Bayer matrix,
so in your TCAM mosaic they're swapped red and blue channels (outer foil is gold not blue).

Posted by: machi Feb 4 2016, 04:05 AM

One of the test of workflow for TCAM data.
Full resolution image of Yutu.

Posted by: jccwrt Feb 5 2016, 01:25 AM

Ooh, thanks for pointing that out! That completely slipped my notice. I've swapped the color channels and replaced the images in my original post with updated versions.


I played around with a different set of stereo images to see if they worked any better, and I think it was a success. Here's an anaglyph looking back at the lander.

https://flic.kr/p/CS1Gw7
https://flic.kr/p/CS1Gw7 by https://www.flickr.com/photos/132160802@N06/, on Flickr


And a quick mosaic using Chang'e 3 Observation 11 data. Unfortunately I can't produce a 360 degree panorama due to severe light streaks in 3 of the images. I do really love the low Sun angle and how it highlights how uneven the surface really is. I reduced the contrast a little so that the lighting isn't so harsh.

https://flic.kr/p/DnqUgy
https://flic.kr/p/DnqUgy by https://www.flickr.com/photos/132160802@N06/, on Flickr

Posted by: Hungry4info Feb 6 2016, 12:13 PM

Was the Chang'e 3 data release incomplete? I remember images such as this one from 21 Dec 2013 that don't seem to be included in this release.
- Edit: I see that was clarified on the planetary.org site.

Posted by: elakdawalla Feb 6 2016, 02:58 PM

Quanzhi Ye told me that these images weren't released all at once, that they appeared over time. I'd encourage people to check in at the data center site from time to time and check the number of posted images to try to notice when new ones appear.

Posted by: Hungry4info Feb 20 2016, 07:33 PM

I'm having trouble accessing https://planetary.s3.amazonaws.com/data/change3/pcam/png/PCAMR-C-009_SCI_N_20140113190843_0008_A_2C.png taken at 2014-01-13 19:08:43. It tells me "This XML file does not appear to have any style information associated with it"

Posted by: elakdawalla Feb 20 2016, 08:17 PM

That is the highly informative error that you get when the file isn't found. I'll look into it Monday.