I processed the Mastcam-Z Left video of the high speed test of April 10th which ended prematurely.

Click to view attachment

Higher resolution version of the video here.

First flight will (might) be Monday morning!
https://www.nasa.gov/press-release/nasa-to-...-soon-as-monday

Thorsten

Droll Doug. I love a good tautology. But there must be a reason why these images are so compressed when other are not. that is what I was getting at. No known reason is I guess the answer I am seeking.

You can see from the filenames that these were jpeg compressed onboard at quality 50. Presumably they didn't want to have the huge downlink volume of so many uncompressed or lower compression video frames, and decided that for this purpose the quality was enough.

When returning realtime-compressed video, 16 1280x720 frames have to fit in 2 Mbytes, so you can do the math about how much compression you can use. https://mastcamz.asu.edu/cameras/tech-specs/

Ingenuity should be airborne right about now.

The Nasa coverage said data will be received any minute now.

success !

That flight sequence was amazing.

Congratulations, and what an image of the shadow and Perseverance tracks!
Edit: and now the two color photos from the rover!

Flight video as seen from Perseverance on Youtube (time 40:55)
https://www.youtube.com/watch?v=p1KolyCqICI

Tweets of the images/video from the livestream:

• Shadow pic
• Video from Perseverance

Haven't noticed a press release yet, maybe we'll not get that until the in-depth presser later?

https://www.youtube.com/watch?v=p1KolyCqICI&t=2309s

has the altimetry plot. The change in altitude to the hover appears to be very close to 3m. The duration of the hover appears to be around 40s if I read the scale correctly. Ascend and descend take less than a few seconds, perhaps one second.

https://www.youtube.com/watch?v=p1KolyCqICI&t=2525s has the image sequence.

Makes sense. There were a lot of very clear images of the airfoils, then these so I figured something was different.

Just watched the Ingenuity first flight full frames from ZL. Incredible. Go download them! The hover and yaw rotation is distinctly visible. This clearly sets the stage for a drone mission with ground support and opens up a lot of the planet for further discovery.

115 seconds.

0672089482.269 to
0672089597.207

(*edit - from the altimetry plot, looks like airfoil rotation started at about 089425 and ended at 089580. Interesting the altimeter seems to have adjusted itself )



A few of the ascent (27 sec) and descent (22 sec) frames are missing. Maybe they are saving them to make the full presser video.

Link to Raw Images of Ingenuity's first flight

*edit2: As you scroll through the video stills of the flight, there are some compression artifacts, of course. These are not entirely random it seems. You can see what appear to be columns of refracted warming atmosphere rising to the top of the image (descending if you reverse the scroll) which are emphasized by the artifacts the compression algorithm produced. If that is the case, it demonstrates a great deal of thermal interaction between the very thin atmosphere and the ground.

Wow for the shadow picture. I'm curious about the 'shutter' speed, and can't find a relevant answer.

Andy

They might answer that at the presser coming up in an hour. Put the question on Twitter with the #MarsHelicopter hashtag. I will too and they might pick it up.

The shutter speed of the Ingenuity camera must be incredibly fast to capture a clear shadow of the airfoils while they rotate 360 degrees in 24 ms (66 μs / degree). If you allow 5 degrees of rotation without visible blurring, that would be .333 ms of shutter open time. 1/3000 s.

30 frames per second for the helicopter nav cam.

A quick mp4 from the frames, cropped and scaled x2:

https://user-images.githubusercontent.com/6...6ffe263906f.mp4

workflow:

get frames from roverpics and wget

import as layers in gimp

use Filters - Animation - Playback to check

crop

export as webp (slow but hopefully higher quality then indexed gif)

for smaller file size use some web converter to convert to mp4


and a small gif for convenience (wait a bit for the spin up), 200ms per frame:

Frame rate ≠ exposure.

https://www.researchgate.net/publication/32...gy_Demonstrator suggests.....

• Navigation (NAV) Camera. This is a global-shutter, nadir pointed grayscale 640 by 480 pixel sensor (Omnivision OV7251) mounted to a Sunny optics module. It has a field-of-view (FOV) of 133 deg (horizontal) by 100 deg (vertical) with an average Instantaneous Field-of-view (IFOV) of 3.6 mRad/pixel, and is capable of acquiring images at 10 frames/sec. Visual features are extracted from the images and tracked from frame to frame to provide a velocity estimate

A useful article is at -

Particularly check the comments - estimated 1/30,000 s:

Digital Photography Article on Ingenuity Apr 19

(May be wrong based on my rough math above - I came up with 1/3000 s)

In the shadow picture, why the heck is the shadow of the body/solar panel and legs black, but the shadow of the rotor blades is a lighter tone? The blades don't appear to be translucent.

John

Mimi just played the entire video showing all the missing frames. Wow, and they are on the Raw server now. It looks like they hit the raw server as fast as they could have.

Watching the press conference now; looks like it was indeed some dust on the panel, that has now been taken off, and is now getting more power than before!

Because the blades are moving at 2500 rpm.

From this thread: https://twitter.com/sdamico/status/1384205372668350465



I was going to suggest that the high blade speed resulted in dark photons falling over the side, but I didn't dare.

Full flight video:

https://youtu.be/wMnOo2zcjXA

I estimated very roughly 1/4000s to 1/6000s based on the new lower-elevation image, which shows the blurring more clearly.

Yeah, that's a short exposure for Mars, but the low resolution of 640x480 will help to keep noise down.

The blade shadow edges are almost sharp, so RPM can't be the primary reason. I'm happy with the additional low-gain integration explanation, though.

John

Ingenuity hovering in the air: navcam anaglyph:
Click to view attachment
and cross-eyed:
Click to view attachment
Very rough colour tweak, otherwise no processing.

Some differences in the ground surrounding Ingenuity are visible after the flight - the ground behind and especially to the right got darker, presumably due to dust blown by the blades. Here's a before-and-after gif:
Click to view attachment
And the stretched difference of the two frames:
Click to view attachment

Congrats to you and team Mike, great work on taking Ingenuity video.
Any chance you'll eventually do a write up on the work that was needed to squeeze this video out of the Mastcam imaging system?
Also, did the project ever request a proposal to develop upgraded electronics for higher frame rates?

Hi John !

Great observation, took me a few seconds to figure out.

remember this was a pretty low-altitude shot, so the rotors are appreciably higher off the ground than the body (box). So, while both items block the direct solar beam, hence forming shadows, the shadow under the box sees a lot less scattered light from the Martian sky (bigger solid angle obstruction for each point) and so is pretty dark. The rotors obstruct a lot less of the sky as seen from each little patch of the shadow, and so the scattered light fills in the rotor shadows more, hence the lower contrast.

Ralph

Hi Ralph-

Thought of that, but the rectangular solar panel is higher still and has a black shadow. So no.

John

This video rate has been a part of the Mastcam design all the way back to MSL (and was used for the MSL MARDI descent video, although there we had the flash space to store the images in raw form and read them out with compression later.) Though I'll acknowledge that there's not a huge amount of detail about it in https://link.springer.com/article/10.1007/s11214-020-00755-x ("Compressed color videos: Bayer pattern interpolated, 8-bit companded, lossy JPEG-compressed images concatenated into 16-frame motion-JPEG GOPs")

If JPL had wanted better video, they could have added another EDLcam I suppose, at some unknown cost delta (putting it on the RSM would have been prohibitive, I suspect, so it would have been fixed-pointing.)

I made animations from cropped raw frames, from most 'raw' (webp) to most compressed (mp4).

webp animation of flight

gif:


mp4 animation of flight

Thanks Mike. I was having trouble figuring out how it got the video data rates over an 8-MBit/s LVDS link, which is the highest link rate mentioned at https://mastcamz.asu.edu/cameras/tech-specs/
However, from "J.F. Bell III et al", I see the link from camera head to DEA can run at up to 120Mbps which would yield a roughly the video frame rate.

Just curious, is the Actel FPGA re-programable?

Another animation (655 frames), slightly smoothed, and color stretched:



as mp4 for full screen:

enh. animation as mp4

workflow:

crop layers in gimp
use gmic to gently smooth all layers
use gmic to convert layers to tiles
color process
convert tiles back to layers
save as .xcf
use ezgif.com to convert to gif, optimize to under 10MB
use ezgif.com to convert to webp, 100% quality, convert to mp4, optionally resize

Thank you (!) Andreas. These GIFs are just great.

As a general question, is there any likelihood of getting clearer uncompressed images down at some point? Being the first flight and all it would be good to have the best possible video of it.

There is no trace of dust at all in the downwash which is remarkable. I guess that was predictable, but still surprising - JPL and the heli-team was also surprised it seems.

When we see so much dust raised in the atmosphere of Mars, it seems there should have been at least some kicked up by Ingenuity on take-off.

This seems to suggest that atmospheric mass has little connection to dust events on Mars. If a highly efficient chopper airfoil revolving 40 times a second over a dust-bed can't raise any dust , it is highly unlikely a gentle Martian breeze could. There is just not enough gas mass to move the dust mass. This experiment has proven that. Yet there are sand dunes and evidence of aeolian erosion all over the place. It's mystifying how all this sand and dust got moved around, and also a puzzle as to how the massive dust events maintain momentum in such a thin atmosphere. Perhaps over a very extended time horizon compared to what we are used to here, millimeters at a time over a billion years.

So dust storms and dust devils seem to be predominantly thermal events, just like thunderstorms on Earth - not that this was unknown, but certainly this video adds weight to the idea that the thermal regime of the atmosphere (vs. atmospheric mass) is the main climate driver on Mars. I always had the idea that a good blow on Mars would create a dust devil or cloud of dust but Ingenuity put paid to that.

Getting video down from Perseverance is already helping us visualize these thermoclines much more precisely as compression artifacts from atmospheric refraction. Could this be a good argument to put better video cameras on future rovers? Better processing power and new codecs (ie. AV1) give us high quality video at lower bitrates so we could perhaps even get HD video clips from Mars at 30 fps in the not too distant future. That is an exciting prospect. The improvements in imaging from MER to MSL to Mars2000 seem to be exponential and is there good reason to expect that to continue?

What do you think Mike? Could a 25 gram 4K cots camera make it into the Mars pipeline anytime soon? We can dream can't we?

Thanks, for me it was also worth learning how to generate such animations from frames (although there may be better tools for that).

The compression artefacts are in the raw png, perhaps a function of necessary local video processing during recording due to storage constraints. Random example:

https://mars.nasa.gov/mars2020-raw-images/p...2_034050J01.png

So the only chance for higher quality may be from still images. Perhaps zcam can record video and still images simultaneously ? Seems unlikely.

I noticed that the timestamps on the images are often 149 ms apart, not 150 ms.

I suppose it is still possible that the launch area just has a thin regolith layer, with the dust fraction having been blasted away during rover landing which was pretty close, and perhaps initial spin tests.

Somebody has to pay for it. EDLcam DDC was 2Kx1500 and 140 grams not counting the processing unit. It's fairly hard to build a camera with decent optics that weighs 25 grams, at least for any definition of decent that I would use. Putting stuff up on the mast is very difficult. Doing things without interfering with the science mission is also difficult. EDLcam didn't have to deal with either of those problems.

Too bad it didn't land near InSight - they could buzz right over and clean her off.

In the raw images - even without processing - there is a large downwash at takeoff and landing. It's subtle - but it's there.

Not really. Sand dunes on Mars do migrate, albeit slowly. This was first demonstrated by MRO in 2011 and Spirit took some nice images of sand ripples migrating during the first dust storm. The ground where Ingenuity was dropped seems pretty dust free. More of a desert pavement. I wonder what the downwash velocity would be with counter rotating blades, given the light weight and low atmospheric density. Not overmuch it would seem.

If its cots, the acquisition cost would be minimal. No doubt the integration cost would be quite high though. I have in mind (most would agree its a pretty decent camera for what it does - likely under 25g) something like:


iPhone 12 Pro Max camera subsystem
.

That could be put on any part of the rover, mast, arm or landing hardware. Perhaps even replace the NavCams. You would just need the processing power to support it which again, could be cots. Ingenuity demonstrates cots components can work in planetary mission environments.

I acknowledge it may not fit within the risk envelope you guys have to work in though. But what amazing science results it would provide.

It wasn't surprizing to everyone!

We've seen wind move dust many times on Mars - that's what kept the MER rovers going, after all!

Is it clear we're not just seeing nothing more than varying compression artifacts, rather than thermal currents in the air?

You would still have to run power and comms cables up to it ( which on the RSM or arm is very very non trivial ) as well as survival heaters.

And then you have to get the data back to Earth - which is also non-trivial.

MCZ doesn't do interframe video processing as such - it just encodes a sequence of still images as regular jpeg files (and bundles them up into groups of 16 frames). But they have to fit those 16 into 2MB (to fit the 8Mbps data rate) which means they have to use heavy jpeg compression (quality 50).

Thanks for the explanation. That explains seeing jpeg compression artefacts in pngs. Presumably the 16 frames get compressed together and are then split out as single frames (in the camera or later).

That makes perfect sense. And I know the imaging teams are pushing to get the maximum performance from the cameras.

My underlying point is that these new requirements (capturing motion either of the atmosphere or the helicopter) ought to motivate getting improved video technology on future missions. We have very capable, incredibly cheap, light and energy efficient camera subsystems available off the shelf now so perhaps those should be looked at for possible integration into future missions. If helicopters are now in the future exploration toolbox, it seems those platforms would be ideal for these types of cameras.

shouldnt this go in a future hardware thread, not the thread on ingenuity operations.

there is a lot of noise getting added with incorrect image analysis/design suggestions that requires corrections from multiple people every time.
mods please delete whatever you decide, its just getting frustrating trying to keep up with actual ingenuity information