Electroholic Anonymous

It would be great if Magic Lantern could repeat their work for the 5D line, that would be heaven. Otherwise, if I find a way to recover the DAT file into a healthy CRM file, I might just use that. It should not be difficult, because ending the 17 min recording in my video, does not appear to take any considerable processing. If I compare the DAT file to the CRM file with a hex editor (HxD), the data looks the same (also in SciPy and Matlab I get the same data entropy readings, of course, it's compressed RAW data after all) The header is different though, what's missing is metadata, how many frames are recorded, block entries for audio and video data, etc. A simple copy over from the header of the healthy CRM to the unfinished DAT file, did not work well in Premiere and Resolve, they recognize it, but Premiere gives frame access exceptions (I forgot the precise error description), and Resolve shows black video. It is not easy (for me), to rebuild the correct metadata for the DAT file. It is not easy for the c200 either because it lasts a long time on that camera to repair a DAT file. Hopefully our friends at Grau expand their program with support for the R5 soon, it support a long list of camera's and formats, I trust they will be able to do it. Their program is able to repair the MP4 files, but the resulting 8K video is all corrupted, although the metadata is correct, correct codec information, number of frames and all. I have send them a healthy and corrupted MP4 too. Thanks Andrew, please keep pushing with the wonderful people of Magic Lantern. Where can we donate to Magic Lantern to get this show going? Crowd funding anyone? If that does not work out, I am confident (and foolish) enough to shoot with the power interruption method, once a reliable DAT file recovery option is available.

I am afraid they will not fix it. There already video production companies who by prosumer equipment instead of de cinema lines. It's a hell of a lot cheaper, and not only that, you can mingle more easily with a GH5 (or a7iii or an R5) then with a c300 or Red. That's the danger, that more of these companies, think, we'll go for the R5 instead, no need for the c300, or c-something-RF. Even worse than not fixing it, Canon might close the loophole(s). Then I rather stay on 1.0.0.firmware and find a way to use the unfinished DAT files. I already sent the power interrupted 8K RAW file (which appears on the CFexpress card with the extension DAT instead of CRM) as wel as a healthy CRM file to grauonline.de. Again, see the decription and first comment of my video in which I describe all of this. The C200 can finish such a DAT file, the R5 does not offer this (with the current firmware). So let's say your R5 crashes, or your battery malfunctions, then your recorded data (DAT file) is unrecoverable at this time. That's where grauonline comes in, they offer a program that can automatically repair video recordings that are corrupted. I hope I hear back from them soon.

\ As is wrote above, despite 60W of Peltier cooling power supplied, with a copper heatspreader covering the whole back of the R5 behind the LCD screen and cooling it to -5 degrees Celsius, risking internal condensation, it has ZERO impact on the time I can record in 8K RAW, with my R5 with firmware v 1.0.0. I was willing to walk around with this ugly contraption if it would give me longer recording times. I need a full NP-F970 battery per hour for the maximum cooling. I doubt Tilta will apply that much cooling power.

What deep freeze experiment? I missed that. I have built a Tilta freezer myself 3 weeks ago, and I can confidently state that based on my experiments, ramming 60W of power into a Peltier element, cooling the back of the R5 to -5 degrees Celsius, that it has ZERO impact on the overheating timer. See pictures of my crude but effective Peltier cooling system. The cooling works, the sensor is 5 to 10 degrees colder, the card slot around 10 degrees Celsius colder, but zero impact, at leat with this version 1.0.0 R5 firmware. The Tilta is a marketing creation, and drawn up by a technical drawing artist. Having worked a considerable part of my career as a hardware and software engineer for a telecom networking giant, this is my guess. I could be wrong, and that Canon and Tilta are working together and that Tilta has access to a newer firmware that does take temperature measurements into account. Good suggestions, I also thought of that. However, I can rule this out based on my experiments. I am describing this in the description and first comment of my video. Sometimes, roughly 1 in 3 times, the card does not write fast enought after a sudden power interruption. The manual of the R5 specifically states to perform a low-level format when this happens, this is indeed effective. In the other cases that the card writes just fine after the power interruption, it gladly gives 15:00 for 8K RAW, every time.

At the end of my video linked above I am showing a neat trick that I found out. If the overheating timer shows 00:00 it is still possible to start recording within a few seconds. Here's how: Power down the R5 normally. Press the record button and keep pressing it the whole time. Switch on the R5, and as soon as it finishes its startup process It will start recording. Once it starts recording you can stop pressing the record button. I am showing this a couple of times at the end of my video linked above. This leads me to believe that the overheating timer was slapped on in the last instance of the development process, at least the timer procedure kicks in very late in the startup process, after reading out user commands apparently. Very odd. The trick is not very useful though, it only records for a couple of seconds to at most a minute. It appears the overheating timers are very precise, the time it records before it shuts down due to 'overheating' correlates with the time since the previous time it shuts down from overheating.

I can confirm Andrew's findings, It works the same in 8K RAW on CFexpress,. I recorded my test session and published it on youtube, you can see how I monitored the temperatures with a thermal camera, and Ninja V, both visible in the video here: https://youtu.be/IagTjRO5Y30 Thanks a lot Andrew! Hopefully we can push Canon to fix this, otherwise let's hope the magnificent folks at Magic Lantern will do their magic again! Thanks for all your hard and honest work.