10 new images releases, arm moved:https://mars.nasa.gov/insight/multimedia/raw-images/?order=sol+desc%2Cdate_taken+desc&per_page=50&page=0&mission=insight
NASA seems to be at some point saving them as jpg before converting them to png.
All onboard IDC/ICC image processing is done by the lander flight software running on the lander computer. After an image is read out from the camera, a shutter image is also acquired and subtracted from the image of interest; this removes frame transfer readout smear and dark current. The raw Bayer image is then demosaicked into RGB triplets, color balanced onboard using preloaded color correction coefficients, and companded to 8 bits using a 12- to-8 bit square root lookup table (LUT). The resultant images are JPEG-compressed by the lander computer and packetized for downlink. Typical JPEG quality values used for deployment activites are 85, 90, and 95, which correspond approximately to compressed bit rates of 1, 2, and 3 bits/pixel, respectively (the exact relationship between compression quality and bit rate is scene-dependent).
Quote from: ugordan on 12/02/2018 06:16 pmNASA seems to be at some point saving them as jpg before converting them to png.That point is on Mars. From this PDF:
Please check your quoting, that's not my quote.
By the way, it looks like lossy compression (though not jpeg) was used on the MER rovers as well: https://en.wikipedia.org/wiki/ICER
Quote from: theinternetftw on 12/02/2018 11:10 pmBy the way, it looks like lossy compression (though not jpeg) was used on the MER rovers as well: https://en.wikipedia.org/wiki/ICERYes, MER used quite an interesting wavelet-based compression. The difference is that those images were by definition monochromatic as the cameras used either rotating filter wheels or a fixed one. MSL was the first such landed mission that used bayer color filters on the chip itself and the high frequency bayer mosaic "noise" made such lossy compression schemes really bad unless you did a debayer already on the spacecraft. Once you did that, might as well use good ole jpeg for the resulting color product. Seems like a bit of a shame they didn't adapt ICER for color and it's as if we regressed back to a less sophisticated jpeg algorithm, but hey - it works.
No reason to spend time adapting algorithms when basic JPEG is dead simple to implement.
Looking at the latest pictures from InSight, it looks like the Context Camera has dropped the lens cover, but there is still dust on the lens, is that correct?
Quote from: niwax on 12/03/2018 07:56 amNo reason to spend time adapting algorithms when basic JPEG is dead simple to implement.Perfectly fine for InSight, I was really commenting about MER->MSL transition. Wavelet should theoretically give slightly better image quality than DCT, but it's perhaps less resilient to data dropouts.
