Are these the real Martian colors? Options

[djellison]

FWIW - accounting for the dust cover - it looks pretty much how Mars has always felt in my head.

This - also - is exactly how it looks in my head. http://mars.jpl.nasa.gov/msl/multimedia/ra...E1_DXXX&s=0

It's a judgement call at the end of the day. Even dragging a single JPG between two monitors that are not calibrated can result in dramatic changes. Ditto printers etc etc.

[ugordan]

Are you talking for public release or for transmission to Earth?

Onboard compression and transmission to Earth.

[RoverDriver]

Onboard compression and transmission to Earth.

In that case AFAIK it is wavelet.

Paolo

[ugordan]

In that case AFAIK it is wavelet.

I checked the MARDI, MAHLI and Mastcam specs, they specify JPEG as lossy with support for motion JPEG video. Don't know about the hazcams and navcams, but since they're b/w I wasn't even thinking about them.

[Deimos]

I checked the MARDI, MAHLI and Mastcam specs, they specify JPEG as lossy with support for motion JPEG video. Don't know about the hazcams and navcams, but since they're b/w I wasn't even thinking about them.

Haz and navcams are indeed the same as MER (wavelet or lossless), with one thumbnail and one downsampling & compression level per exposure. MMMcams are jpeg or lossless or raw, with one thumbnail, no downsampling, and the ability to recompress & transmit out of flash when needed.

[fredk]

I can tell you that color IS SUBJECTIVE

As someone who's studied electromagnetism, it's always irked me to some degree to hear people talk about the subjectivity of colour. As far as perception of colour is concerned, I agree completely.

But there is a well-defined sense in which colour is completely objective. A display could reproduce the full spectrum and intensity of visible light that would be present (as a function of direction) a couple of metres above somewhere on the Martian surface, at some time during the day. (I don't care too much where or when.) Of course ordinary rgb displays could never do that. But in principle it could be done, to whatever precision you want. That is what I think about (dream about) when I wonder what colour the Martian surface is. I don't care about the subjectivity of perception, or the fact that the display might not cover the full 4pi steradians ("all the way around") and the light from around the edges of the display may affect my perception, or the fact that my perception of the colour will change over time as I get acustomed to the orangeness.

I don't care about any of that. I just want to see that spectrum!

[RoverDriver]

I checked the MARDI, MAHLI and Mastcam specs, they specify JPEG as lossy with support for motion JPEG video. Don't know about the hazcams and navcams, but since they're b/w I wasn't even thinking about them.

Ah, sorry, it is my perspective that considers only engineering cameras. ;-)

Paolo

[siravan]

I checked the MARDI, MAHLI and Mastcam specs, they specify JPEG as lossy with support for motion JPEG video. Don't know about the hazcams and navcams, but since they're b/w I wasn't even thinking about them.

JPEG 2000 is wavelet based (in contrast to the original JPEG which was DCT based). Based on this article http://boingboing.net/2012/08/06/mars-curi...r-boing-bo.html, hazcams uses a JPEG-2000 lite algorithm.

[ugordan]

Yes, yes, I know about hazcams (or indeed all cameras of MER heritage) using wavelet, I was talking about the new, color cameras provided by MSSS.

[siravan]

There is a sentence in the Mastcam description http://www.lpi.usra.edu/meetings/lpsc2010/pdf/1123.pdf that implies it uses wavelet instead of DCT. It says "A typical image is likely to be 1152 x 1536 pixels owing to JPEG encoding constraints". Now, DCT uses 8x8 blocks and does not have this limitation. On the hand, JPEG 2000 usually uses 64x64 or 128x128 blocks (note that 1152=64*18 and 1536=64*24).

[ugordan]

I think that's rather some hardware/software constraint on the size of a JPEG-encoded frame, but mcaplinger can undoubtedly set us straight.

[Deimos]

Note that Mastcam thumbnails are 8x8 times smaller than the full frames they represent. And are themselves compressed. DCT is perfectly happy with multiples of 64, for obvious reasons.

[Malmer]

As someone who's studied electromagnetism, it's always irked me to some degree to hear people talk about the subjectivity of colour. As far as perception of colour is concerned, I agree completely.

But there is a well-defined sense in which colour is completely objective.

...

...I just want to see that spectrum!

Exactly the way I think of it. Perception is by its very nature subjective. the incoming stimuli that triggers that perception is objective and quantifiable.

...you just need some really magical tech to do it

you would need at least 20000*20000 pixels per eye in stereo. The RGB approximation is not enough by a longshot so you would need a much higher spectral resolution. (perhaps 20 channels or so to capture most spectral detail) you might need to capture polarization aswell. This data would have to be in HDR with at least 16 stops of dynamic range.

And given that the eye-brain perception machine is inherently temporal in nature so you might aswell throw in that you would need the data as a massive lightfield that you can view with full 6 axis head tracking in at least 200Hz with extremely low latency.

that should give you a pretty indistinguishable sensation of looking out over the vistas of mars.

until then Im pretty happy with what i can get from the MER/MSL combo


/Mattias

[dvandorn]

This discussion reminds me very much of one of the appendices to the hardcover volume containing a number of the science results papers from Mars Pathfinder. The appendix discussed the color of Mars.

Because of the various super-res pans they ran through each of the filters, they had a good number of color images, and the appendix followed the discussion as to how to best present the colors.

It then went on to present, as closely as they could possibly manage, the color, contrast and lighting of the scene as it would appear to a human eye standing on the surface of Mars. They then adjusted the image to how it would appear under the same pink sky but with sunlight as bright as that on Earth (which is the mix they tended to use for press release images), then under white light (i.e., pure white sunlight and also a pure white sky, no tint at all). And finally they adjusted the image to how it would appear literally on Earth, with the admixture of scattered blue light from our blue sky.

It was a very interesting presentation, which has from then on given me a mind's eye image of Mars as a rather flat-toned, low-contrast place with a feeling of overcast even under a cloudless sky.

-the other Doug

[mcaplinger]

MMM uses plain vanilla JPEG with either 4:4:4 or 4:2:2 chroma subsampling (and grayscale luminance only for the narrowband filters.) The dimensional limits referred to only apply to constraints of image size set by the hardware compressor (and only apply if you want the thumbnail to be exactly 8x smaller with no dropped pixels.)

The main advantage that we have for "true color" is that the Bayer color responses are similar to the eye's, rather than being the square narrowband RGB filters used on MER. In my experience with sunlight illumination you really don't have to do any white balancing and the out-of-camera JPEG color is pretty accurate. See http://www.msss.com/images/science/mastcam...nyona19i_cc.jpg, which was brightened very slightly but looks fairly accurate without any processing.