IPB

Welcome Guest ( Log In | Register )

36 Pages V  « < 29 30 31 32 33 > »   
Reply to this topicStart new topic
MSL Images & Cameras, technical discussions of images, image processing and cameras
Deimos
post Oct 8 2014, 12:50 AM
Post #451


Martian Photographer
***

Group: Members
Posts: 352
Joined: 3-March 05
Member No.: 183



QUOTE (wildespace @ Oct 7 2014, 04:36 PM) *
...shooting in raw captures the scene and its colours "as is" (if we ignore the camera's specific colour response nuances).


'Shooting in the raw' captures *only* the camera's specific response to its environment, mostly the product of illumination and intrinsic color.

So I would not ignore camera response. But I am biased that way. There is no magic 'as-is' that is revealed by patterned microfilters sitting on top of a semiconductor. If you want to learn about an element of a system, you need to isolate it from other elements: Mostly, what you get is response convolved with illumination * color. And, if you want a geologist to find the result informative, what you want is a whole different response convolved with a different illumination times the same intrinsic color (and you want to get it from a whole range of displays with their own responses). That is what radiometric calibration and white balancing tries to achieve (or at least it's one relevant goal).

You seem to want the illumination * color product, which is fine. White balancing (on PDS, in a point-and-shoot, whatever) tries to do something else (also fine). But 'raw' is still not magic, and the information provided by mcaplinger is a good orientation, and the early-mission images of things of known color are a good check.

For environmental color (i.e., illumination * subject), see also:
http://marswatch.astro.cornell.edu/pancam_...projects_1.html
http://marswatch.astro.cornell.edu/Bell_etal_SkyColor_06.pdf
Go to the top of the page
 
+Quote Post
JohnVV
post Oct 8 2014, 03:05 AM
Post #452


Member
***

Group: Members
Posts: 890
Joined: 18-November 08
Member No.: 4489



Non scientific but ...
I have stated this before but for new people .In the 80's and 90's i worked in Photographic darkrooms
by the 2000's i was replaced by digital . There were not to many still using real FILM

-- from the RAW jpg site
http://mars.jpl.nasa.gov/msl-raw-images/ms...787E01_DXXX.jpg

on average i have found that for the JPG's
--- pds corrected jpg
http://pds-imaging.jpl.nasa.gov/data/msl/M...787C00_DRCL.JPG

i find that dropping the GREEN by 5 ( Magentia +5 ) and adding 5 red is darn close
this is how i would adjusted the color
[attachment=33917:0126.jpg]
Go to the top of the page
 
+Quote Post
mcaplinger
post Oct 8 2014, 05:18 AM
Post #453


Senior Member
****

Group: Members
Posts: 2504
Joined: 13-September 05
Member No.: 497



Everyone's subjective opinion about what the images should look like notwithstanding, the PDS corrected products represent the best way we know of to remove instrument signature. If you want more insight into that, I've tried to exhaustively document the radiometric properties of the cameras on the PDS volumes in the file MSL_MMM_CAL.TXT on each PDS volume (for example http://pds-imaging.jpl.nasa.gov/data/msl/M...MSL_MMM_CAL.TXT )


--------------------
Disclaimer: This post is based on public information only. Any opinions are my own.
Go to the top of the page
 
+Quote Post
James Sorenson
post Oct 8 2014, 05:30 AM
Post #454


Member
***

Group: Members
Posts: 691
Joined: 21-December 07
From: Clatskanie, Oregon
Member No.: 3988



And I can't express more how much it is a treat to have one of the actual guys here that helped build and calibrate the camera's. He and his team are the true experts. The best data available to help answer the question "What color does Mars look like?" is in the PDS, use it people. smile.gif


Go to the top of the page
 
+Quote Post
Gerald
post Oct 8 2014, 12:12 PM
Post #455


Senior Member
****

Group: Members
Posts: 2346
Joined: 7-December 12
Member No.: 6780



QUOTE (mcaplinger @ Oct 8 2014, 07:18 AM) *
... I've tried to exhaustively document the radiometric properties of the cameras on the PDS volumes in the file MSL_MMM_CAL.TXT on each PDS volume (for example http://pds-imaging.jpl.nasa.gov/data/msl/M...MSL_MMM_CAL.TXT )

I've been looking closely at these spectra a while ago. There has been one entry which I wasn't sure how to interprete:
QUOTE
*** Filter 3 (800nm)

wavelength (nm) transmission
--------------- ------------
201 0.2066

Is this first entry representative for the range between 200 nm and 790 nm, a single data point, or is it a typo?
Go to the top of the page
 
+Quote Post
ugordan
post Oct 8 2014, 01:49 PM
Post #456


Senior Member
****

Group: Members
Posts: 3648
Joined: 1-October 05
From: Croatia
Member No.: 523



QUOTE (James Sorenson @ Oct 8 2014, 07:30 AM) *
The best data available to help answer the question "What color does Mars look like?" is in the PDS, use it people. smile.gif

James, IMHO using the RDR DRXX straight up as an example of camera raw color is misleading since you're apparently not correcting for the fact that product has a linear brightness-to-DN encoding, whereas all the other products you show are gamma-corrected "out of the box" (the raw JPGs not strictly being 2.2 gamma, but the square root encoding of the source data brings it close to this).

DRXX is typical of radiometric products for other missions and is the reason why I've been harping about the importance of gamma correction for spacecraft images. Bland outer solar system targets look more "cool" without it because the contrast is stronger and colors are more saturated, but it doesn't mean that's actually the way the bodies look like. Looking at the surface of Mars in this representation helps bring that point home. Here's how that DRXX product looks at my end when I take into account the data is linear and then convert it to sRGB colorspace:

Attached Image


--------------------
Go to the top of the page
 
+Quote Post
mcaplinger
post Oct 10 2014, 03:57 AM
Post #457


Senior Member
****

Group: Members
Posts: 2504
Joined: 13-September 05
Member No.: 497



QUOTE (Gerald @ Oct 8 2014, 05:12 AM) *
Is this first entry representative for the range between 200 nm and 790 nm, a single data point, or is it a typo?

We have narrowband filter data from 200 to 1100 nm but if the filter is only 50 nm wide there's not much point in showing all the out-of-band values (which are all down at the noise floor), so I stripped them out to make the document shorter, and this is a line that I missed.


--------------------
Disclaimer: This post is based on public information only. Any opinions are my own.
Go to the top of the page
 
+Quote Post
elakdawalla
post Oct 27 2014, 08:13 PM
Post #458


Administrator
****

Group: Admin
Posts: 5172
Joined: 4-August 05
From: Pasadena, CA, USA, Earth
Member No.: 454



I've been writing the MAHLI section in my book and have learned something about scaling MAHLI images that I figured might be of interest to some of you. To determine the scale of a MAHLI image, you need to know the working distance between MAHLI and the target. It's linear, so it's easy: once you know that number, you calculate scale by this relationship:

pixel width (microns) = 6.9001 + 3.5201 * working distance (cm)

The problem is finding out the working distance. In the MAHLI image headers, you will see notes in the RATIONALE_DESC field referring to working distances of, most commonly, 1 or 2 or 5 or 30 cm. But these are rover planner notes and are measured from a different baseline than what the MAHLI team means. Rover planners measure working distance from the minimum safe distance, which is about 2 millimeters beyond the 17-millimeter-long contact prongs that project from the front of MAHLI to protect the instrument. So you need to add 1.9 cm to the working distance in the RATIONALE_DESC field to get the approximate working distance to plug in to the above equation.

Of course, the real distance between the camera and the target will not be exactly that for a variety of reasons. It doesn't really matter for the more distant standoffs like 30 cm, but when MAHLI is working super close to a target, a difference in working distance of a few millimeters can make a difference in pixel scale. Now, none of us probably needs to be so precise, but once I started going down this rabbit hole I couldn't stop. If you want the greatest accuracy in pixel scale in MAHLI images, you can determine it from the motor count of the focus mechanism, which is also provided in the header, under FOCUS_POSITION_COUNT. Here is the empirically derived equation for determining working distance (w) of the in-focus part of a MAHLI image from the motor count (m):

Attached Image


The equation is in a Yingst et al. abstract called "Cameras on Landed Payload Robotic Arms — MAHLI on Mars and Lessons Learned from One
Mars Year of Operations", to be presented to the Second International Workshop on Instrumentation for Planetary Missions, November 4, 2014.


--------------------
My website - My Patreon - @elakdawalla on Twitter - Please support unmannedspaceflight.com by donating here.
Go to the top of the page
 
+Quote Post
Gerald
post Dec 5 2014, 12:54 AM
Post #459


Senior Member
****

Group: Members
Posts: 2346
Joined: 7-December 12
Member No.: 6780



A ruler based on the formulas in the paper:
Attached Image

Note, that for focus-stacked MAHLI images the motor count for the first image doesn't necessarily correspond to the best-focused single image.
Go to the top of the page
 
+Quote Post
PaulH51
post Feb 25 2015, 12:41 PM
Post #460


Senior Member
****

Group: Members
Posts: 2425
Joined: 30-January 13
From: Penang, Malaysia.
Member No.: 6853



Looking for advice :

We have recently seen the majority of the full frame mast camera images being issued by JPL as Bayer files. Does anyone know if this is a temporary issue, or if we will see this continue?

The reason I ask is I am not happy with the results of de-bayering these images using ImageJ and its add-in, as I am losing a degree of the image quality as well as increasing the green cast. If JPL are likely to continue issuing MC full frames in this bayered format, then I will seek another 'open source' windows application that can de-bayer without the issues I am getting the way I am using ImageJ.

Here is a link to a 'stitch' using 16 frames of de-bayered sol 903 images (no corrections applied) LINK

Any advice on how JPL are likely to issue these images in the future would be greatly appreciated

TIA
Go to the top of the page
 
+Quote Post
Gerald
post Feb 25 2015, 01:34 PM
Post #461


Senior Member
****

Group: Members
Posts: 2346
Joined: 7-December 12
Member No.: 6780



Hi Paul,
that's not really an issue. The Bayer-encoded images are the lossless images as they come out of the camera ccd. But after jpg compression as in the published edr jpegs some of the information is lost, such that full-quality images can only be generated from the PDS versions.
For de-bayering consistent in color with the compressed images you need to apply the correct de-bayering 5x5 matrices as implemented in the camera firmware.
These needed matrices are published in the PDS (appendix D, p.228 of the document MSL_MMM_EDR_RDR_DPSIS.PDF).
The remaining green cast can be removed by applying the proper color correction weights for each of the rgb components.

The 2048 pixels version of your pano, after color correction (without new de-bayering), square root encoded, as a comparison for color correction:
Go to the top of the page
 
+Quote Post
PaulH51
post Feb 25 2015, 02:27 PM
Post #462


Senior Member
****

Group: Members
Posts: 2425
Joined: 30-January 13
From: Penang, Malaysia.
Member No.: 6853



QUOTE (Gerald @ Feb 25 2015, 09:34 PM) *
Hi Paul,
that's not really an issue. The Bayer-encoded images are the lossless images as they come out of the camera ccd.

Hi Gerald, looks like a little reading is required, many thanks for the pointers
Go to the top of the page
 
+Quote Post
mcaplinger
post Feb 25 2015, 03:40 PM
Post #463


Senior Member
****

Group: Members
Posts: 2504
Joined: 13-September 05
Member No.: 497



QUOTE (Gerald @ Feb 25 2015, 06:34 AM) *
For de-bayering consistent in color with the compressed images you need to apply the correct de-bayering 5x5 matrices as implemented in the camera firmware.

I don't think what interpolation kernel you use has much to do with the color, but the 5x5 preserves the image sharpness a little better. Of course once through the additional JPEG compression there's no telling what may have happened to the Bayer data; if it were up to me I would never present the data in this format.

Paul's images look perfectly respectable to me, perhaps a little off in color.

BTW, I'm surprised that lossless compression is being used as much as it is -- I'd have thought that taking more images with JPEG would be preferred, although we may be returning many images in both JPEG and lossless.


--------------------
Disclaimer: This post is based on public information only. Any opinions are my own.
Go to the top of the page
 
+Quote Post
fredk
post Feb 25 2015, 04:28 PM
Post #464


Senior Member
****

Group: Members
Posts: 4246
Joined: 17-January 05
Member No.: 152



Having the lossless Bayer-encoded images will definitely be better once they appear on the PDS. In the meantime, the jpegging introduces artifacts that mess up the Bayer pattern and typically introduce little green blocks. There have been several attempts to clean those up (I tried Fourier filtering) - see the posts starting around here. But the best results I recall seeing were from Ant using gimp - see this post.

Anyone else who's had good results can chime in...
Go to the top of the page
 
+Quote Post
PaulH51
post Mar 12 2015, 10:41 AM
Post #465


Senior Member
****

Group: Members
Posts: 2425
Joined: 30-January 13
From: Penang, Malaysia.
Member No.: 6853



QUOTE (fredk @ Feb 26 2015, 12:28 AM) *
Having the lossless Bayer-encoded images will definitely be better once they appear on the PDS...

I was hoping that JPL would have reverted to the normal JPEG images, but the compressed bayered images remain the norm for panorama sets, with occasional colour jpegs for specific targets. i tried GIMP for debayering, but I am still unable to recreate the level of 'finish' I used to get.... Am continuing to seek more understanding, so I can slowly improve my skill set.

I am not complaining about JPL, as I am sure they have their reasons for releasing the jpeg'ed bayer images onto the public servers, but the lack of readily available full frame colour mast camera images, could eventually put a sizeable dent into their reputation for great outreach. Time will tell....
Go to the top of the page
 
+Quote Post

36 Pages V  « < 29 30 31 32 33 > » 
Reply to this topicStart new topic

 



RSS Lo-Fi Version Time is now: 28th March 2024 - 04:19 PM
RULES AND GUIDELINES
Please read the Forum Rules and Guidelines before posting.

IMAGE COPYRIGHT
Images posted on UnmannedSpaceflight.com may be copyrighted. Do not reproduce without permission. Read here for further information on space images and copyright.

OPINIONS AND MODERATION
Opinions expressed on UnmannedSpaceflight.com are those of the individual posters and do not necessarily reflect the opinions of UnmannedSpaceflight.com or The Planetary Society. The all-volunteer UnmannedSpaceflight.com moderation team is wholly independent of The Planetary Society. The Planetary Society has no influence over decisions made by the UnmannedSpaceflight.com moderators.
SUPPORT THE FORUM
Unmannedspaceflight.com is funded by the Planetary Society. Please consider supporting our work and many other projects by donating to the Society or becoming a member.