Left/right pairs from both rovers and both sources (JPL and Exploratoriun) shows a systematic difference of jpeg files dimension in favour of right version (tipically 5% on NavCam/PanCam pairs). See example.
Moreover, you If you look carefully to images, you will find, there is a clear difference in terms of fine details: Click to view attachment
I do not know if this strange issue is due to different jpeg compression ratio on the server or is more serious (lens difference? extremely improbable!).
Other explainations??? (Maybe people often dowloading .img uncompressed images can answer... )

I noticed this from the very first Sols, but apparently no one highlighted it here and most people continue to use left images to make stitch or other considerations... viceversa, I reccomend to use always right version, when possible!!!

I never noticed it myself but you might find something that explains the bias on the Bell MER Athena Pancam Investigation document or in the MER/Pancam Data Processing User's Guide, the latter is a fascinating explanation of how to interpret the data.

There are obvious differences between the Left and right cameras (different filters, slight calibration differences, and the different offselts from the Pancam Mast) but I can't find anything that indicates left images would be smaller than right.

Thanks, helvick, very interesting articles!
(the right Bell MERdocument url is this one).
The issue we are talking about cannot arise from filters difference, because it affect also the Navigation cameras... however, in the second document I found an interesting difference: "Due to the confined space on the camera bar, the left Pancam is rotated 180 degrees relative to the right Pancam. The images stored in the EDRs, however, have already been rotated" [...] "As a result, effects such as transfer smear (see below) appear to be in opposite directions in the left and right images."
Can this difference induce a smear variation in the two sensors and also different "smooth" effect from application of the shutter-smear removal algorithm?
If yes, also .img untouched images should exhibit the difference!...

The compression used onboard the rover might explain the difference, though I'm not sure why images from the left camera would consistently be compressed at a different rate than the right. If you are talking about pancam, make sure to compare apples with apples though (L2 and R2, L7 and R1) , as it *is* standard practice to compress one filter differently than another filter.

If you do want to check and see though, some values are available within the PDS tags which may answer this question, which are described at : http://pds-geosciences.wustl.edu/geodata/m...amera_dpsis.pdf

INST_CMPRS_QUALITY: A JPEG or ICER specific variable which identifies the resultant or targeted image quality index for on-board data compression.
INST_CMPRS_RATE Provides the average number of bits needed to represent a pixel for an on-board compressed image.
INST_CMPRS_RATIO Provides the ratio of the size, in bytes, of the original uncompressed data file to its compressed form.
INST_CMPRS_SEGMENT_QUALITY Quality level for each ICER segment. For MER, the quality values are used to determine whether a decompressed image was losslessly or lossy compressed. If the segment quality value was zero, then that segment was losslessly compressed. Any non-zero value indicates lossy compression for that segment. If all the segment quality values were zero, then the entire image was losslessly compressed.

There are other compression-related parameters that descend deeper into the detailed working of ICER and LOCO also available within the "COMPRESSION_PARMS" group.

Based on reactions in another thread, I realized that very few people noted this important highlight.
The ussue still clearly visible and, finally, someone else re-discovered it! Look at these NavCam details taken on Sol896 by Opportunity:
Click to view attachment
The left image is dramatically more noisy than right one, especially on uniform regions (sky). This cannot be due to higher jpeg compression because (contrary to my first highlights) in this case left image original size is even bigger than right one! In fact, even by removing jpeg artifacts the left image still worse, as showed in this sharpened view:
Click to view attachment

I didn't checked the calibrated images; so, again, can someone more expert with the PDS site check if this difference still visible there? This control could be an hint also understanding the possible cause.
In the meantime, pls, use as much as possible right Nav/Pan images for finest detailsanalysis/stitches... (unfortunately, we cannot use for true color compositions due to the lack of corresponding filters on the right PanCam! )

Is the same true of the RAD's out of the PDS?

Might be interesting to do a comparison of the JPG, and then all the various PDS products, for left and right of one observation.

Doug

I am no image expert, so take this a pure guess.

Maybe the drivers/scientists need only one sharp image for details and another one (showing almost the same area) for the stereo effect with less detail. So they compress the left more than the right and save some downlink volume. One sharper image to start with, another for the stereo effect. Isn't the most sharpest filter on the left pancam? So they compress this less. And with navcam they decided once to do it the same way (by convention).

To check for this someone should compare uncompressed 13 filter images of selected science targets, if there are any of these (should be very rare, RAT holes?). If they are uncompressed, they should have the same size.

Analyst

I've emailed JB. Navcam's not his dept - but I'll ask anyway.

I have always noticed the right cam images are better than the left ones.

I'm embarrassed. I've never noticed this, although I sometimes work with stereo pairs. Out of habit, I almost always look at the left image and rarely the right as single images. I guess this started when working with color images, as in L-2-4-7 and carried over to the right Navcam and Hazcam images.

--Bill

I think there is a pretty simple reason why people have favored left-camera images: the visible-light wavelength filters of Pancam are on the left camera. Well, that's why I've favored the left cameras, anyway. For MMB panoramas I favored the left Navcam because it matches up better with the left Pancam. At least, I assume that is the case - it's been a long time since I've checked.

I honestly don't see that much different in quality between the two cameras except for the sky portion of the images. Even in the sky portion of the images, it seems like the noise that is visible in the left navcam is just smoothed out a bit in the right. I am very curious as to what is the difference and why, though. I hope JB can tell us something there.

I think Analyst's explaination is very plausible, even if there is an error in the post: the sharper filter is on the RIGHT Panoramic camera and is the R1 (436nm) - remember the WOW! effect when they finally decided to use it for Victoria Crater some weeks ago?
The check suggested with uncompressed full filter images of selected science targets should be decisive (unless we have an answer from JB!)

Thanks dilo. I always thought the empty "filter" L1 is the sharpest, but probably it is only the most light sensitive. But anyway, this could explain it. Jim Bell will know.

Analyst

Edit: Deleted mistake.

Analyst, in my knowdlege, the R1 is a narrow-ban blue filter in the Panoramic Camera.
You are probably confusing with the unfiltered NavCam.
See here.

L7 and R1 are the deep blue / violet that has excellent detail.

Doug

Dilo, this may seem counterintuitive, but it actually does make sense! If you convert two images, identical except one with more noise, into jpegs, the jpeg compressor has to work "harder" to capture all the noise in the noisier one, and the result is a bigger file. That's what we're seeing here.

This is another good reason to use R navcams: the filesize is smaller!

Well, even though I started this resurgence in the L vs R topic, I'm embarrased that I never noticed the effect until earlier this year. (Time to start reading the imagery forum regularly!)

I checked several of the jpl navcam jpegs. The effect is very prominent on all Oppy navcam pairs I looked at; a good example is this left navcam and the corresponding R navcam. (Ahh, those were great days, huh? What would that heatshield look like...)

I never saw nearly as big a difference with Spirit. For example, this and this.

Also, this L navcam noise doesn't look like ICER compression noise. Here's an example of an image with obvious compression artifacts.

Next, I went ahead and looked at some pds images. Here's the above sol 115 Oppy L navcam view; top is jpl site jpg, bottom is mrd pds:
Click to view attachment
Here are the R views (jpeg top, mrd pds bottom):
Click to view attachment

The result: The L pds is far cleaner than the L jpeg. Curiously the R pds has a slightly dirtier look than the R jpeg!

It appears that there's something about the way the Oppy L navcam jpegs are processed on earth that introduces extra noise. The noise definitely does not originate with the rover cameras.

Excellent work, fredk. Really!
Looking to images you published (sol 115 Oppy navcam view) the quality of uncompressed L image is impressive, not only compared to the corresponding L jpeg (which appear really bad) but also compared to the "dirtier" uncompressed R version!
Now we can reasonably say that the main cause is related to jpeg compression made on Earth for Web publication. We do not know the true reason for the difference, but paradoxicalally could be related to higher noise in the R original images, which are somehow less/better compressed. Not so strange, if we thing that, in the digital audio domain, a "dither" technique is often used to improve ADC/DAC performances through addition of small random noise signal (below 1 lsb)...
This is only an hypothesis, I think we are all waiting definitive explaination from Jim Bell!

If nothing else, maybe we can get JB to persuade JPL to "clean up" their public-release JPEGs fof the left channel.

--Bill

I'm glad this left/right thing was mentioned. I can't believe I hadn't noticed it before. I suspect it is because it is most noticeable in the sky, and I devote most of my concentration on the rocks and soil. That's my excuse, anyway. At first I was thinking it was only apparent on the sky portions of images, but after looking at some navcam pairs at 200-300% magnification, I am convinced there are more and larger artifacts in all parts of the left navcam images.