Search results

AFAIK, Digic is an application-specific integrated circuit (ASIC). You are right, it almost certainly has a general purpose processor of the "embedded kind," capable of running ARM instruction sets. Not sure how fully-fledged it is; do you know what core they use? In my work, we tend to...

Cameras dont have cpus, they have microprocessors. Of course they transcode in hardware; that hw generates heat... What I said last post, paraphrased.

I can totally see this happening but 'low power' is also relative. When Canon designed the Digic X with its design power (I'd estimate it somewhere around 5W based on power consumption), they must have realized very early on that thermals will be challenging. In this case, engineers tend to put...

No, a microchip runs tens or more degrees hotter than its ambient, especially when doing something like transcoding 8K. Also, you cannot protect the silicon by by measuring ambient; you need the thermal sensor on the die (where exactly is a nontrivial question). Maybe..

From a technical perspective, this is getting more confusing with every new bit of information. Two temperature sensors on imaging sensor assembly (and none in the Digic X) suggests that the microprocessor does not overheat at all. However if true, internal vs external recording should have...

Genuine question, what features has a 7D that are not covered with a R6?

Possible with optics, i.e. a RF to EF-M speed booster design.

Fully agree but I'd expect that 8K will be mainly a production tool and 4K mastering will remain "standard" for quite some time. Resolution is really getting to a point of diminishing returns. Unless screen sizes increase dramatically (at equal viewing distances), I don't see many tangible...

So true. I think they had to spend a lot of time on this. It is probably one of the Digic X's crown jewels (encode 8K30 10-bit 422 on a microprocessor with a few Watt thermal envelope). Software encoding of 8K in real-time is only possible with the highest-end CPU, e.g. 64core Threadripper...

Me too; apparently there is an 8bit 8K30 mode..

How do you actually do pixel binning with CMOS sensors? Is there any ILC manufacturer that does this? I can only think of the following two ways: - Quad-bayer sensor (like smartphone cameras): combine adjacent analog photo sites (this would need a complete rewrite of Canon's internal demosaic +...

Do you happen to know how non-HQ samples? Is it some form of pixel-binning or line-skipping?

Good question. In HW GPU encoding (I use nvenc), I never noticed a relevant difference between h264/h265 encoding, which seems to be confirmed by https://www.pugetsystems.com/labs/articles/Premiere-Pro-14-2-H-264-H-265-Hardware-Encoding-Performance-1778/#RawBenchmarkResults This does not mean...

H265 was designed to achieve about half the data rate of H264 with similar apparent quality. Real-time SW encoding of 8K30 or 4K120 is currently not feasible for anything but the most powerful CPUs (with 225W+ power envelope)...

(1) I'm not backing any horse. I'm trying to explain to a (possibly brainwashed) person why the things likely are the way they are. (2) If you think that any company has 10x margin in their competitive product, you lack common sense; especially if you are talking about the company that just...

(1) eoshd is a drama queen ;) (2) interesting (3) unfortunately pointless unless they can match their findings with a 3D thermal CFD model of the camera and know exactly where the thermal sensor is (it is usually not in the hot spot)

I hear this quite a bit but always wondered why. Can you explain? Initially, I always thought this was due to computation power but most GPUs today support h255 encode/decode (except for 4:2:2 chroma subsampling that lacks support independent from codec and bitrate)...

This is of course not an exact science but I'll give it a try: Let us assume that when electronic devices are off, the firmware is not active (and no internal heat is generated)*. Then, the constant recovery time is physics: temperatures reach equilibrium with an exponential function with time...

Canon has always been slow in adopting new HDMI standards (see 5Div). R5 is almost certainly HDMI2.0 since the advanced_user_manual states that HDMI-output of 8K-movies yields 4K-movies. Hence, I don't think we'll see 8K HDMI output unless they decide to implement 8bit 420 in general. That code...

Recovery times are purely passive, i.e. how fast the internals can move the heat to the outside. This is a HW limitation and Cannot really be addressed with firmware (except for maybe allowing higher internal temperatures in general).