Canon R5 - Officially safe to use 2022?

[Zack Birlew]

Hey, everybody, so now that the unlocked firmware has been released for the Canon R5, is it officially safe to use it as an 8K digital cinema option? While the R5C is just a bit more, the savings for the R5 could be a nice windfall if there's no problem with the overheating issue now but has anybody experienced any issues like sensor damage or memory card instability when used for long form video?

 

[AtticusLake]

I have no idea what "the unlocked firmware" is... could you post a link to that?

I have tested the R5 with firmware version 1.6.0, and it works pretty well. The new "Auto power off temp" setting certainly lets it run for longer. In my tests, in a living room at about 19-20 °C, I actually couldn't get it to shut down, even recording for 3-4 hours continuous. In 4kHQ mode, it seemed really stable -- no warnings at all. In 8k mode, there were intermittent warnings, and it seemed like it was pretty close to the edge, though it did record for 4 hours continuous with no shutdown.

Note that "continuous" has one huge caveat -- it has a 30-minute record time limit. So every 30 minutes you get a break in your recording. If you only record clips less than 30 minutes, no problem.

Bottom line, I think you're fine as long as it's not too hot. If you're outdoors in the sun, you may have issues, certainly with 8k.

BTW, personally, I would not (and do not) record 8k -- 4k HQ gives you basically the same resolution, but with much smaller files and less heat. (8k is 8k photosites, not pixels. 4kHQ is 4k pixels.)

I wrote up my tests here: https://moonblink.info/MudLake/gear/r516

And you can see my thoughts on the camera after a year of use here: https://moonblink.info/MudLake/gear#R5

 

[Zack Birlew]

What I meant by the "unlocked firmware" was that clearly Canon locked the camera up with a problematic internal cooldown timer so now with firmware 1.6, the camera is effectively "unlocked". That's good to hear that it works just fine for continuous recording, at least to a point, so has anybody put it through long form tests with multiple take recordings for narrative filmmaking? Features? Shorts?

 

[AtticusLake]

Just FYI, the cooldown timer was removed in version 1.1, so what you're seeing out there is bias, not facts. In fact for a long time the R5's heat performance has been pretty good; in 1.6 they just made it better, essentially by allowing it to get hotter. Canon made some people very unhappy, because of how they released the camera; but if you're looking to buy, you should be looking for factual information on how it fits your needs, not people venting their anger.

So your question is the right one. Sorry I can't help you with it, since I haven't been doing narrative film-making. I have been using it heavily for shooting landscape movies, typically running 15-30 minutes per clip, outdoors, summer and winter, in the Scottish highlands. So, not terribly hot -- not exactly Tucson. OTOH this was almost all on firmware 1.4, and I never saw an overheat. Add to that the tests I linked to above, and that's what I have for you. Hope it helps.

 

[Clermond]

For video I'd always choose the R5c over the R5 as the R5c has many benefits
- all codecs are HQ, the R5 has only HQ in 4K24/25/30
- all non Log codecs are 10bit. The R5 has only 10bit in Clog/Clog3
- R5c has the hot shoe for the Tascam Audio Adapter
- with more space due to the fan the display can be fully articulated. I find this a massive downside of the R5 when you plug in an HDMI or USB-C cable

I find the benefits are worth the difference of about 630€ net.

edit
I own two R5, one R5c, one C300iii

 

[AtticusLake]

I also have an R5 and an R5C, and I also generally use the R5C for video these days. But not for exactly the same reasons.

- all codecs are HQ, the R5 has only HQ in 4K24/25/30

Sadly this is not really true. One nice thing about the R5 is that it explicitly tells you which modes are HQ; with the R5C, you have to guess. So I ran some tests on a res chart. (the numbers are only relative.) The R5 scores 10.5 in HQ, 5.5 in LQ. The R5C scores 10.5 up to 30 FPS; 8.0 at 60 FPS. However it scores 13.0 in MP4 8k. But only 5 at 120FPS. So it's better than the R5, but don't expect the same resolutions at all frame rates.

Your other points are all good -- particularly the articulating display, which is really nice. (Personally, I always shoot log.)

But the big reason I use the R5C is the video-centric operating system. This doesn't affect the results much -- though 8k at 60FPS could be cool -- but mostly it's just a lot nicer to use for video. You have shutter angle, waveform, false colour, and of course no 30-minute limit.

I find the benefits are worth the difference of about 630€ net.

For video shooting, I would agree.

 

[Clermond]

I
Sadly this is not really true. One nice thing about the R5 is that it explicitly tells you which modes are HQ; with the R5C, you have to guess. So I ran some tests on a res chart. (the numbers are only relative.) The R5 scores 10.5 in HQ, 5.5 in LQ. The R5C scores 10.5 up to 30 FPS; 8.0 at 60 FPS. However it scores 13.0 in MP4 8k. But only 5 at 120FPS. So it's better than the R5, but don't expect the same resolutions at all frame rates.

For my understanding HQ is the definition for downsampled images not by line skipping.

 

[AtticusLake]

For my understanding HQ is the definition for downsampled images not by line skipping.

Right. This is usually called "oversampling", and it gives you a significantly better image than line skipping or pixel binning. This is particularly visible when there could be aliasing or moire, like in patterns of fine lines.

Based on my chart tests, I would say that both the R5 and R5C have oversampling up to 30 FPS. But beyond 30, the R5C produces a lower resolution. It doesn't look as bad as the R5's line skipping, but not as good as full oversampling; so I'm not sure what's really going on there.

This isn't too surprising. Video is a huge amount of data, and oversampling is a lot of computation. So it's no surprise that when you double the frame rate -- and hence the amount of data -- it starts taking shortcuts.