|
The full landing sequence of #Chandrayaan3 I wrote a python script to undo the frame blending from the original video and then upscaled the images with AI. Credit: #ISRO / Simeon Schmauß 12 comments
The full video sequence was discovered by @Astro_Aure@twitter.com on the ISRO website: https://www.isro.gov.in/media_isro/video/Home_video/Ch3_lander_descent_rover_rollout.webm You can see the original frame blending interpolation in that video. Unfortunately @isro doesn't provide a high quality version of the images... To find the original unblended frames my python script uses a high-pass filter to look for edges in the frame. Basically, the fewer edges it finds, the more likely it is to be an unblended frame.  Equally unfortunately, that isro website "cannot be found" when I click on the link here in the US. 🤷♀️ @SubtleBlade @janetlogan It should work now. the "www" of the url was lost while copying the text from twitter and some websites don't work without it. FWIW: that's a sysadmin error. It is entirely possible to auto redirect the DNS to resolve "www" when it's omitted from the request. So not on you really. 👍 @SubtleBlade @janetlogan I should have known better though because I already encountered this earlier today with the same url 😅 Normally one wouldn't really encounter this because you just copy a complete url. As we've all learned by now, the network formerly known as Twitter thinks rules and conventions are for everyone else. As does it's current owner, apparently. @stim3on beautiful!! Great job. Really shows how involved a landing like this is. … idle thought … How hard would it be to overlay an altitude and maybe a speed display on there? A realtime clock might be nice too. @sb Thanks! Unfortunately ISRO is very slow to release stuff on the mission. We only have a few shots of the telemetry screen from the livestream, nothing really concrete. https://www.youtube.com/live/DLA_64yz8Ss?si=q0pUYBz38iX8v1Lk&t=1290 https://www.youtube.com/live/DLA_64yz8Ss?si=4TQ0FYnNN4-_ZuLw&t=1731 |
Gregory's Server
My goal was to extract clean images so I can interpolate the whole video with AI to get a very smooth video like @TJ_Cooney@twitter.com did for the rover deployment sequence.
Unfortunately there is too much movement in the video so the interpolation didn't produce good results. If anyone wants to give this video another try I'm happy to provide the extracted frames.
https://twitter.com/TJ_Cooney/status/1695251767615783374