3 ms·

Does anyone at NASA consider that these practices are deceptive? Remember they also add false color, and I don't think I've ever seen the manipulation process n

Does anyone at NASA consider that these practices are deceptive? Remember they also add false color, and I don't think I've ever seen the manipulation process noted. At times I've looked for true images, and it's been difficult to find out if any image is 'real' or manipulated. Imagine if a newspaper published photos like that; it would be seen as deceit. Personally, it makes me feel that NASA's public announcements are not trustworthy, that they have a sophisticated PR machine and we're not really seeing the truth.

tzahola 8y ago
Space probes’ cameras are not optimized for pretty pictures. They collect scientific data that oftentimes falls outside of the visible spectrum. For PR material they obviously synthesize visible images from this data, as the layperson appreciates pretty pictures with false colors, as opposed to six or more images with grayscale intensities at various wavelengths.
mturmon 8y ago
You have to be more careful about what you mean by "these practices". False-color images, in which some quantity you care about (e.g., water vapor, [1]) is projected into a map or spatial coordinate, are pervasive throughout the sciences. Nobody would seriously contemplate putting some kind of interpretive warning on such images, because we all know you can't see water vapor or ocean salinity directly. From such non-deceptive cases, you have a slippery slope. After you work with spatially-mapped imagery for a while, it's natural to insert (invisible) infrared channels of an earth-observing satellite as the red channel of an RGB image. Similarly, it's natural to color-code certain spectral bands of solar images, in extreme UV, as a shortcut to highlight flares, even though you can't come close to seeing these bands directly [3] [4]. It's also natural to adjust the contrast or related properties like white balance, of planetary images. After all, your own monitor is adjusting the contrast through a gamma correction [2] that isn't well-standardized. I think people have complained about these kinds of enhancements before -- in some cases, with reason -- so now you do in fact see disclaimers about contrast enhancement on NASA images. [5] If you pay closer attention, I think you will notice these disclaimers, too. (I did not cherry-pick [5], I just looked at the JPL news feed until I came to a news release that might involve a real-world scene - the next example I saw was [6], which also has a disclaimer). I haven't been involved in that many NASA press releases, although I have done work with the investigator mentioned in [1]. Of course, for a press release, you have to simplify complex issues and make analogies and all that stuff. Hopefully the science comes through, and also hopefully, there is a link to the published paper which has the technical details. That last is the key, because it offers some guarantee that the result was vetted, and a method to review the original data. The amount of care that goes in to getting calibrated data is hard to convey. It's critically important, and all the science done at NASA and with NASA data depends on it. For instance, just think for a bit about all the geometric and photometric calibration that had to be done to mosaic up the little self-portrait image of Curiosity. They had to back out the joint angles of all the actuators from the camera all the way back to the robot, and color-correct for different exposures, and rescale to account for distance differences, and account for parallax with a camera lens model (which itself had to be calibrated) and .... And that's just for a non-science image. It doesn't really need to be photometrically calibrated. The actual science images would have much more careful localization and photometric calibration. A person I work with routinely cross-calibrates in-situ instrument footprints of size ~1mm onto this imagery to get the correspondence between mineral grains (visible in camera images) and mineralogy information. For instance, the widely-used CAHVOR(E) geometric camera model that maps pixel indexes (i,j) in the CCD array, to physical locations in space, was invented in 1978 at JPL [7]. And things have come a long way since then. As a sample of the documentation of the MAHLI camera that took those images, see [8]. That paper was followed with an even more detailed post-landing calibration document [9] - about 100 pages, condensed from years of R&D and calibration. What is my point? First, it is always a stretch to map the data from a special sensor like that back to an RGB image. And second, if you really do want to understand the relation of what you're seeing back to the original physical setup (i.e., learn the "truth" as revealed by the measurement), you have to do a lot of geometric/photometric analysis. And third, rather astonishingly, that analysis is also done routinely, and published for whoever wants to read/critique/improve. [1] https://www.jpl.nasa.gov/spaceimages/details.php?id=pia21209 https://www.jpl.nasa.gov/spaceimages/details.php?id=pia21209 [2] https://en.wikipedia.org/wiki/Gamma_correction#Windows,_Mac,_sRGB_and_TV/video_standard_gammas https://en.wikipedia.org/wiki/Gamma_correction#Windows,_Mac,... [3] https://www.nasa.gov/content/goddard/how-sdo-sees-the-sun https://www.nasa.gov/content/goddard/how-sdo-sees-the-sun [4] https://www.nasa.gov/content/goddard/sdo-aia-171-angstrom/ https://www.nasa.gov/content/goddard/sdo-aia-171-angstrom/ [5] https://www.jpl.nasa.gov/news/news.php?feature=7137 https://www.jpl.nasa.gov/news/news.php?feature=7137 [6] https://www.jpl.nasa.gov/news/news.php?feature=7118 https://www.jpl.nasa.gov/news/news.php?feature=7118 [7] https://www.sciencedirect.com/science/article/pii/0146664X78901120 https://www.sciencedirect.com/science/article/pii/0146664X78... [8] https://link.springer.com/article/10.1007/s11214-012-9910-4 https://link.springer.com/article/10.1007/s11214-012-9910-4 [9] https://www.researchgate.net/publication/278728737_Curiosity%27s_robotic_arm-mounted_Mars_Hand_Lens_Imager_MAHLI_Characterization_and_calibration_status?channel=doi&linkId=55849c9208ae7bc2f448403c&showFulltext=true https://www.researchgate.net/publication/278728737_Curiosity...
- TeMPOraL 8y ago
  Thanks for the detailed explanations. I wish space science put a little bit more focus on trying to produce true-color images to the best extent possible. Scientists themselves might be primarily interested in the scientific data, but everyone is also interested in knowing how those things look. That is, if we were to look at them with our own eyes, what would we see? The danger of unlabeled false-color maps that look "natural" is that they'll later find their way into arts - movies, videogames, posters, book covers, etc.
- forapurpose 8y ago
  Wow, thank you for such a informative response. I know that in any field high precision and accuracy generally are very expensive (something about the first 90% being cheap, and the last 10% ...), so I expected it would be a complex process - especially given the hyper-constrained resources on Mars. But wow - sounds fun. HN is not the usual audience, of course; most people don't think to ask the question. And of course, maybe others republish the images without noting the alterations. And it seems my criticism/question is outdated, as you say NASA has made more effort to address this issue. > Nobody would seriously contemplate putting some kind of interpretive warning on such images, because we all know you can't see water vapor or ocean salinity directly. It's hard to for experts to underestimate what 'we all know' (I and my peers have the same problem in my field - look at what HN users think typical computer users know, can grasp, and will do); looking at that map[0] I could easily imagine many people thinking it was somehow real. Here's a recent story, no exaggeration: I was watching a sports commentary show widely praised as among the most intelligent, and they had on a guest who revealed that he had studied astrophysics in a prior life. One of the commentators asked (and I'm paraphrasing, but this was really the question): Maybe you can settle a argument for (another commentator) and me: What is further from New York, Los Angeles or the Moon? Because I can see the Moon from New York, but I can't see Los Angeles. [0[ https://www.jpl.nasa.gov/spaceimages/details.php?id=pia21209 https://www.jpl.nasa.gov/spaceimages/details.php?id=pia21209
caymanjim 8y ago
They are not deceptive in the least. Skepticism is fine, but your question comes from a fundamental lack of understanding. As an example, when you see a pretty picture of Mars in a magazine, it's composed of dozens or hundreds of individual images, captured as long, thin image strips, as a satellite orbits the planet. Those images come in looking absolutely nothing like the final product. You're viewing a rectangle, but the source data represents a strip along the surface of a sphere. It needs to be rectified, aligned to dozen of other such strips, and projected on to a flat surface. Unless you're projecting these images back on to a sphere, there's a whole lot of stretching and alignment going on. The images are taken at different times of the day, during different seasons. Software needs to balance the albedo so that it doesn't look like a Picasso painting. Most NASA images are false color, because they are not taken with normal optical cameras. Much of the good science comes from imaging outside the human visual spectrum. The THEMIS instrument on Mars Odyssey images 10 infrared spectral bands, none of which can be seen with the human eye. Other instruments record ultraviolet light. There is no way for humans to see any of this data, so they pick three spectral bands and assign them to red, green, and blue so that humans can see something. Different band selections produce different images of the same area and serve to emphasize different minerals or other natural materials. Science is done on all bands of the data, but pictures help provide context. The raw data for all these images are freely available to anyone, but asking to see the "real" image before it's been processed is a meaningless request. There is no such thing as a "true image".

tzahola 8y ago

Space probes’ cameras are not optimized for pretty pictures. They collect scientific data that oftentimes falls outside of the visible spectrum. For PR material they obviously synthesize visible images from this data, as the layperson appreciates pretty pictures with false colors, as opposed to six or more images with grayscale intensities at various wavelengths.

mturmon 8y ago

You have to be more careful about what you mean by "these practices". False-color images, in which some quantity you care about (e.g., water vapor, [1]) is projected into a map or spatial coordinate, are pervasive throughout the sciences. Nobody would seriously contemplate putting some kind of interpretive warning on such images, because we all know you can't see water vapor or ocean salinity directly. From such non-deceptive cases, you have a slippery slope. After you work with spatially-mapped imagery for a while, it's natural to insert (invisible) infrared channels of an earth-observing satellite as the red channel of an RGB image. Similarly, it's natural to color-code certain spectral bands of solar images, in extreme UV, as a shortcut to highlight flares, even though you can't come close to seeing these bands directly [3] [4]. It's also natural to adjust the contrast or related properties like white balance, of planetary images. After all, your own monitor is adjusting the contrast through a gamma correction [2] that isn't well-standardized. I think people have complained about these kinds of enhancements before -- in some cases, with reason -- so now you do in fact see disclaimers about contrast enhancement on NASA images. [5] If you pay closer attention, I think you will notice these disclaimers, too. (I did not cherry-pick [5], I just looked at the JPL news feed until I came to a news release that might involve a real-world scene - the next example I saw was [6], which also has a disclaimer). I haven't been involved in that many NASA press releases, although I have done work with the investigator mentioned in [1]. Of course, for a press release, you have to simplify complex issues and make analogies and all that stuff. Hopefully the science comes through, and also hopefully, there is a link to the published paper which has the technical details. That last is the key, because it offers some guarantee that the result was vetted, and a method to review the original data. The amount of care that goes in to getting calibrated data is hard to convey. It's critically important, and all the science done at NASA and with NASA data depends on it. For instance, just think for a bit about all the geometric and photometric calibration that had to be done to mosaic up the little self-portrait image of Curiosity. They had to back out the joint angles of all the actuators from the camera all the way back to the robot, and color-correct for different exposures, and rescale to account for distance differences, and account for parallax with a camera lens model (which itself had to be calibrated) and .... And that's just for a non-science image. It doesn't really need to be photometrically calibrated. The actual science images would have much more careful localization and photometric calibration. A person I work with routinely cross-calibrates in-situ instrument footprints of size ~1mm onto this imagery to get the correspondence between mineral grains (visible in camera images) and mineralogy information. For instance, the widely-used CAHVOR(E) geometric camera model that maps pixel indexes (i,j) in the CCD array, to physical locations in space, was invented in 1978 at JPL [7]. And things have come a long way since then. As a sample of the documentation of the MAHLI camera that took those images, see [8]. That paper was followed with an even more detailed post-landing calibration document [9] - about 100 pages, condensed from years of R&D and calibration. What is my point? First, it is always a stretch to map the data from a special sensor like that back to an RGB image. And second, if you really do want to understand the relation of what you're seeing back to the original physical setup (i.e., learn the "truth" as revealed by the measurement), you have to do a lot of geometric/photometric analysis. And third, rather astonishingly, that analysis is also done routinely, and published for whoever wants to read/critique/improve. [1] https://www.jpl.nasa.gov/spaceimages/details.php?id=pia21209 https://www.jpl.nasa.gov/spaceimages/details.php?id=pia21209 [2] https://en.wikipedia.org/wiki/Gamma_correction#Windows,_Mac,_sRGB_and_TV/video_standard_gammas https://en.wikipedia.org/wiki/Gamma_correction#Windows,_Mac,... [3] https://www.nasa.gov/content/goddard/how-sdo-sees-the-sun https://www.nasa.gov/content/goddard/how-sdo-sees-the-sun [4] https://www.nasa.gov/content/goddard/sdo-aia-171-angstrom/ https://www.nasa.gov/content/goddard/sdo-aia-171-angstrom/ [5] https://www.jpl.nasa.gov/news/news.php?feature=7137 https://www.jpl.nasa.gov/news/news.php?feature=7137 [6] https://www.jpl.nasa.gov/news/news.php?feature=7118 https://www.jpl.nasa.gov/news/news.php?feature=7118 [7] https://www.sciencedirect.com/science/article/pii/0146664X78901120 https://www.sciencedirect.com/science/article/pii/0146664X78... [8] https://link.springer.com/article/10.1007/s11214-012-9910-4 https://link.springer.com/article/10.1007/s11214-012-9910-4 [9] https://www.researchgate.net/publication/278728737_Curiosity%27s_robotic_arm-mounted_Mars_Hand_Lens_Imager_MAHLI_Characterization_and_calibration_status?channel=doi&linkId=55849c9208ae7bc2f448403c&showFulltext=true https://www.researchgate.net/publication/278728737_Curiosity...

TeMPOraL 8y ago
Thanks for the detailed explanations. I wish space science put a little bit more focus on trying to produce true-color images to the best extent possible. Scientists themselves might be primarily interested in the scientific data, but everyone is also interested in knowing how those things look. That is, if we were to look at them with our own eyes, what would we see? The danger of unlabeled false-color maps that look "natural" is that they'll later find their way into arts - movies, videogames, posters, book covers, etc.
forapurpose 8y ago
Wow, thank you for such a informative response. I know that in any field high precision and accuracy generally are very expensive (something about the first 90% being cheap, and the last 10% ...), so I expected it would be a complex process - especially given the hyper-constrained resources on Mars. But wow - sounds fun. HN is not the usual audience, of course; most people don't think to ask the question. And of course, maybe others republish the images without noting the alterations. And it seems my criticism/question is outdated, as you say NASA has made more effort to address this issue. > Nobody would seriously contemplate putting some kind of interpretive warning on such images, because we all know you can't see water vapor or ocean salinity directly. It's hard to for experts to underestimate what 'we all know' (I and my peers have the same problem in my field - look at what HN users think typical computer users know, can grasp, and will do); looking at that map[0] I could easily imagine many people thinking it was somehow real. Here's a recent story, no exaggeration: I was watching a sports commentary show widely praised as among the most intelligent, and they had on a guest who revealed that he had studied astrophysics in a prior life. One of the commentators asked (and I'm paraphrasing, but this was really the question): Maybe you can settle a argument for (another commentator) and me: What is further from New York, Los Angeles or the Moon? Because I can see the Moon from New York, but I can't see Los Angeles. [0[ https://www.jpl.nasa.gov/spaceimages/details.php?id=pia21209 https://www.jpl.nasa.gov/spaceimages/details.php?id=pia21209

caymanjim 8y ago

They are not deceptive in the least. Skepticism is fine, but your question comes from a fundamental lack of understanding. As an example, when you see a pretty picture of Mars in a magazine, it's composed of dozens or hundreds of individual images, captured as long, thin image strips, as a satellite orbits the planet. Those images come in looking absolutely nothing like the final product. You're viewing a rectangle, but the source data represents a strip along the surface of a sphere. It needs to be rectified, aligned to dozen of other such strips, and projected on to a flat surface. Unless you're projecting these images back on to a sphere, there's a whole lot of stretching and alignment going on. The images are taken at different times of the day, during different seasons. Software needs to balance the albedo so that it doesn't look like a Picasso painting. Most NASA images are false color, because they are not taken with normal optical cameras. Much of the good science comes from imaging outside the human visual spectrum. The THEMIS instrument on Mars Odyssey images 10 infrared spectral bands, none of which can be seen with the human eye. Other instruments record ultraviolet light. There is no way for humans to see any of this data, so they pick three spectral bands and assign them to red, green, and blue so that humans can see something. Different band selections produce different images of the same area and serve to emphasize different minerals or other natural materials. Science is done on all bands of the data, but pictures help provide context. The raw data for all these images are freely available to anyone, but asking to see the "real" image before it's been processed is a meaningless request. There is no such thing as a "true image".