Canon USA drops new teaser (FX30 competitor?)

[Kino]

6 hours ago, Django said:

I'd say 60p video at 16:9 and the 40 fps max stills burst aren’t directly related. The camera drops to 7K 30 fps in 3:2, and stills use a separate pipeline with buffer/processing limits, so max burst is capped by the stills system, not video frame rate.

That said, the difference in DR between FF and S35 mode could also come from the readout: fewer pixels in S35 can reduce noise, though it’s also possible Canon is doing something in the background we don’t fully know. Either way, it’s a clever and welcome addition to the camera!

I was simply agreeing with ND64's statement that the C50 most likely has 12-bit video readout in standard FF 16x9. There is a connection between electronic shutter limitations and video readout speed, although the S35 sensor crop can allow for higher bit depth or higher readout speeds. Moreover, cropping into S35 will not reduce noise. It will have the oppositive effect. The sensor is also not dual gain like the one in the Panasonic S1II and doesn't offer any of those features. The only way you are getting more DR out of a sensor crop is to switch to a higher bit depth (13 or 14 bit). If you compare the C50 with the R5C, which has 12-bit video readout, you see almost identical latitude (4 stops over and 3 stops under) as demonstrated in the CVP review:

I will admit that the C50 has a beautifully sleek design, and may be worth picking up for the open-gate mode alone, but I don't see much in the way of image improvements over the R5C that I already own. I guess it is a camera made for those who don't have an R5C or those who may need another compact body. In that case, however, I would rather go with a camera that has IBIS, high ISO performance for low light, and greater DR in highlights and shadows. These are qualities I miss in the R5C and that have apparently not been addressed in the C50.

[Django]

41 minutes ago, Kino said:

I was simply agreeing with ND64's statement that the C50 most likely has 12-bit video readout in standard FF 16x9. There is a connection between electronic shutter limitations and video readout speed, although the S35 sensor crop can allow for higher bit depth or higher readout speeds. Moreover, cropping into S35 will not reduce noise. It will have the oppositive effect. The sensor is also not dual gain like the one in the Panasonic S1II and doesn't offer any of those features. The only way you are getting more DR out of a sensor crop is to switch to a higher bit depth (13 or 14 bit). If you compare the C50 with the R5C, which has 12-bit video readout, you see almost identical latitude (4 stops over and 3 stops under) as demonstrated in the CVP review:

I will admit that the C50 has a beautifully sleek design, and may be worth picking up for the open-gate mode alone, but I don't see much in the way of image improvements over the R5C that I already own. I guess it is a camera made for those who don't have an R5C or those who may need another compact body. In that case, however, I would rather go with a camera that has IBIS, high ISO performance for low light, and greater DR in highlights and shadows. These are qualities I miss in the R5C and that have apparently not been addressed in the C50.

Fair points, but I think you’re overstating the certainty here. None of us actually know if Canon is switching bit depth between FF and S35 on the C50. Canon hasn’t published that detail, so it’s speculation. What we do know is Canon are claiming higher DR in S35 mode, so whether that comes from pipeline tweaks, bit depth, or something else, the end result is what matters to shooters.

Also, the CVP review actually concludes the opposite of “no improvements”: they specifically say the C50 has better DR, better AF, C-Log2, the XLR top handle, and a range of new photo/video features the R5C doesn’t have.

For me, the 7K open gate, C-Log2, higher dual base ISO range, dual format recording, cine-oriented form factor, and XLR top handle are what seal the deal over the R5C. Neither are perfect cameras, but the C50 checks enough boxes for me to seriously consider it. 

[ND64]

Its all speculation, but it makes sense to switch from 12bit to 14bit in Super35 mode. The main reason almost all cameras are at 12bit in full frame video is basically power consumption. But in crop modes, they have to deal with less resolution, so the total consumption stays low. There is also another theory. Canon is known for applying NR to the raw video. Maybe the CPU is not powerful enough to apply the same amount of NR applied to Super35, to the 7k image, or if it does, there will be overheating/power consumption issues. Or maybe we're reading too much between the line and the official spec table has incorrect number and will be edited soon.

[eatstoomuchjam]

It could also have to do with readout speed.  14-bit is slower to read out and tends to give worse rolling shutter numbers.  If Canon decided they wanted to hold RS at 15ms, they could probably read out the full sensor at 12-bit, but still have <15ms when reading out the smaller 5K crop at 14-bit.

It seems like a weird thing to do, but more choices are always welcome.

[Django]

12 hours ago, ND64 said:

Its all speculation, but it makes sense to switch from 12bit to 14bit in Super35 mode. The main reason almost all cameras are at 12bit in full frame video is basically power consumption. But in crop modes, they have to deal with less resolution, so the total consumption stays low. There is also another theory. Canon is known for applying NR to the raw video. Maybe the CPU is not powerful enough to apply the same amount of NR applied to Super35, to the 7k image, or if it does, there will be overheating/power consumption issues. Or maybe we're reading too much between the line and the official spec table has incorrect number and will be edited soon.

It's definitely not a typo! Several C50 reviews mention it and there is a B&H C50 talk with the DP of the short film I linked earlier and a senior Canon rep that confirms S35 has a stop more of DR, when pressed on how they achieved that he says its readout & ISO related. That seems to suggest Canon is using the reduced data load in the crop to drive the sensor differently, maybe a slower more precise readout path, or shifting the base ISO behavior to lower noise. Not unlike how Panasonic’s DR Boost works, though probably different sensor readout trickery. 

[Django]

Here is a link to the video: https://www.youtube.com/live/-xwAllKdSW0?si=idF9ndnB6oNJjzKb

It's an interesting talk, the DP mentions how the camera opened up lens choices for their short, from high end Panavision anamorphics to FF, S35 and even S16 glass. In that sense it really behaves like three cameras in one. Sure, the R5C also has crop modes, but the C50 adds open gate, all the desqueeze factors, and a rugged screw mount cinema build. You can also record log master on one card and baked in LUT on the second. That makes it feel like a purpose designed tool rather than a hybrid you have to adapt around.

[j_one]

So maybe it could be the case that the c50 in super35 could get in the ballpark of the c70 without the need of DGO sensor…thoughts?

[ND64]

6 hours ago, j_one said:

So maybe it could be the case that the c50 in super35 could get in the ballpark of the c70 without the need of DGO sensor…thoughts?

I think C70 will still have more latitude to play with. But honestly people are way too much obsessed with these numbers. Past 12 stop, we either digging for random shot noise, or struggle to defeat sun. This shot by Roger Deakins would be rejected by YouTubers, in their mission impossible of saving highlights:

 

[Kino]

The CVP test shows the C50 with a slight advantage over the R5C in the highlights (1/3-1/2 stop), but a disadvantage in the underexposure latitude, which is much noisier. On balance, this new 32mp sensor is basically equal to the R5C in the latitude test despite claims of "improved DR" at the end of the video, a claim that is never demonstrated as they don't use a Xyla chart. After all, they are retailers trying to sell you a new product!

Now I'm sure someone will come along with Xyla charts and show that this camera has a DR advantage over the R5C and is closer to the R5 II. Clog-2 implementation would require that. In the meantime, I'm simply not that impressed with the C50 image or with any of the newer Canon "cinema cameras" for that matter. After the class-leading DGO sensor in the C70/C300 III, it seems that they are going backwards while the competition (e.g., Blackmagic, RED, Nikon, Fuji, Panasonic) is charging ahead . . .

[eatstoomuchjam]

If "improved DR" means 1/3-1/2 of a stop, I'd be inclined to not worry about that, regardless.  But if the S35 mode gets a full extra stop (plus the 1/3-1/2), that's likely to be noticeable.

I'd stick with the standout feature of this camera for many modern shooters being the 3:2 open gate recording option.  That isn't a thing I care about so I'll continue waiting for Canon to release the camera that makes me think it'd be worth selling my R5.  Hint to Canon, it probably involves the letters d, g, and o.  😅