[X] My Assistant

[CosmicRocker]

I'm still trying to figure out a number of things about the new images we are trying to work with. Assuming others are likewise trying to learn, I thought I would open this thread to create a place for such discussions.

I'd like to start out with a comment about raw image contrast. There have been several postings in the main threads about whether or not the MSL raw images have been stretched like those from the MER missions. I am certainly no expert on this, but it looks to me as if the MSL images have not been stretched at all. I haven't tried to analyze all of the image types, but the hazcams and navcams have pixel brightness histograms that are very different from their MER counterparts.

This attached image compares MER and MSL navcams along with their luminosity histograms.


The MSL images clearly are not using the entire, available range of brightness values, whereas the MER raws do. For this reason, the MSL raw images can usually be nicely enhanced by simply stretching the distribution of brightness across the full 256 value range.

[fredk]

My FFT Fourier filtering is also having serious trouble with the solar shot. One of the other techniques used here will probably work better. Here's my best (quick and dirty) shot, using Fourier filtering at 2, 4, and 8 pixel periods, followed by 2px Gaussian blur. No deBayering at all, since the image uses a monochromatic filter anyway.

You can easily see the asymmetric profile (which you could with MER pancam too).

ADMIN NOTE: A bunch of posts that followed this one, discussing image compression etc, were moved to the MSL Images and Cameras thread.

[mhoward]

But all of the smudginess is artifacts - the sun should be very smooth.

I've noticed that they're really compressing the living daylights out of MSL images before putting them on the web - much more so than the MER images are compressed. Or are we talking about a different kind of artifacting.

[mcaplinger]

I've noticed that they're really compressing the living daylights out of MSL images before putting them on the web...

I think they're just using a fixed quality (75, maybe?). Of course it doesn't help that there is sometimes a decompress/recompress and that they JPEG-compress Bayer-pattern data. The final compression ratio seems to be around 8:1 to 9:1. Are the MER images really a lot lower compression ratio?

For the 100mm sun image, since the filter cuts out all of the pixels besides blue anyway, you'd be better off just tossing the other Bayer positions and then upsampling as desired. But it's still going to be a round slightly-fuzzy circle with a bite out of one side.

[mhoward]

Are the MER images really a lot lower compression ratio?

I've been looking at mostly Navcam images so far, so I guess I can't and shouldn't try to say anything about Mastcam. But for MER Navcam images, the JPGs on the web usually are about 200K or more in size; MSL Navcam JPGs on the web seem to be usually around 100K or less. And there just seem to be more and stronger JPG artifacts, at least in images I've looked at of the deck. That could be factors other than compression for the web; for all I know they're being compressed more heavily on the rover in some cases. All I really know is, trying to take a MER Navcam image and make it about the same file size as the MSL Navcam images we're seeing on the web, I have to use a JPEG 'quality' setting of about 40%, which is high compression.

Here's what I typically see (random section of the image that happens to be in front of me):

Attached thumbnail(s)

 

[mcaplinger]

But for MER Navcam images, the JPGs on the web usually are about 200K or more in size; MSL Navcam JPGs on the web seem to be usually around 100K or less.

Well, that speaks for itself. Assuming full 1024x1024 frames, 200K would be about 5:1 and 100K would be about 10:1. I did a quick spot check of some recent Navcams and they were more like 120K, but close enough.

I don't know how they chose the JPEG quality for MER and I don't know how they chose it for MSL, but I would think that 10:1 would be about quality 50 and 5:1 would be about quality 75. My own personal opinion is that 75 would be a more appropriate choice, but nobody's asking me.

As for your example, at that scale they both look pretty crummy (the MSL one has more JPEG artifacts, clearly), but I don't think zooming the image is really a fair test. That said, I wouldn't pick a fight with anyone who says the web release images are compressed too much, but I'm not sure I would use the phrase "compressing the living daylights out of" -- YMMV.

[mhoward]

As for your example, at that scale they both look pretty crummy (the MSL one has more JPEG artifacts, clearly), but I don't think zooming the image is really a fair test. That said, I wouldn't pick a fight with anyone who says the web release images are compressed too much, but I'm not sure I would use the phrase "compressing the living daylights out of" -- YMMV.

Fair enough. Here's a more typically-sized comparison, as good as I can make it at the moment. The MSL version still seems lossy to me. But if it is compression that's being done on the rover to get the images back faster, I'm perfectly happy with that - I'm just so happy to be seeing images at all. I only bring it up because I care.

Normally I would JPEG-compress these images, but well, this is a special case for obvious reasons.

Attached thumbnail(s)

 

[mcaplinger]

The MSL version still seems lossy to me.

At this point I'm tempted to give into my bias and say that all Navcam images look bad to me. (Sorry, Justin. Not all, but some.)

I've always been a bit surprised that the MER Navcams were so grainy-looking. This might be dark current from the relatively long readout time. They're a bit blurry but what can you do with only four elements and fixed focus?

As for wavelet compression, having ICER artifacts interact with JPEG artifacts isn't going to improve the images.

But to recap -- are the public release images on MSL being compressed on the ground more than for MER? Could well be.