Canon R5 Announced with 8k

[AndreeOnline]

Can anyone tell me why canon 5D magic Lantern RAW footage has so much richer color than anything else? Is it a consequence of its limited dynamic range?

Are you saying it's inherently richer, and not a stylistic choice? Do you feel that you couldn't get the same color by grading other footage?

Can you give an example of one rich clip and one that you feel is... I don't know... saturated in the "wrong way"?

 

[Guest]

Are you saying it's inherently richer, and not a stylistic choice? Do you feel that you couldn't get the same color by grading other footage?

Can you give an example of one rich clip and one that you feel is... I don't know... saturated in the "wrong way"?

I told you you were confusing people about DR.

 

[ahalpert]

Are you saying it's inherently richer, and not a stylistic choice? Do you feel that you couldn't get the same color by grading other footage?

Can you give an example of one rich clip and one that you feel is... I don't know... saturated in the "wrong way"?

I'm saying that I don't think if you saturated another camera that its color would look as lush.

2 good ones:
https://www.youtube.com/watch?v=jKMtJPFWkaM
https://www.youtube.com/watch?v=0SHrn9XgEvg

Here's one where I don't like the grade:
https://www.youtube.com/watch?time_continue=42&v=4eFwOnKYeq4&feature=emb_title

But even then I like the way the colors look.

This one looks boring to me probably due to overexposure: https://www.youtube.com/watch?v=a11m0EeKZHA

But I've never seen ML RAW footage that had thin-looking color.

I thought perhaps it has more color depth. Like transparency vs print film.

 

[AndreeOnline]

Well, there is a connection between luminance and saturation. And that makes sense.

There might be something in the ML workflow (I guess everyone used more or less the same tools?), that leads to the development of a certain look? I never did that hack myself.

But my thinking is that it should be possible to achieve that feel since we can control all aspects of the signal, including boosting color separate from luminance.

 

[ahalpert]

Or this one where i don't really like the grade and it's mostly desaturated but the color still looks rich: https://youtu.be/2KwhIAuAIcs

 

[NorBro]

IMO, there is nothing rich or saturated about any of those videos. You can achieve the same quality/look from many different cameras.

The only thing that makes it special is the camera it's coming from and how it's being produced.

If interested, check out this person's channel. A great source to watch a bunch of camera IQ one after another:

https://www.youtube.com/user/powerofcelica/videos

 

[ahalpert]

The color doesn't have to be saturated to be rich.

One of the things that impresses me about ML RAW is that I like basically everything I see of it, whereas I see a lot of crappy looking stuff from any other camera, RED and ARRI included.

That person's channel does indeed look nice and clearly nice grading. But even so, sometimes the red is a little loud. There's a shot from the 1DC video with different colored kayaks that look a little flat in tone. Sometimes it lacks subtlety.

A lot of subtle tone variation from ML RAW. Maybe I'm deluding myself, but that's just how I feel. I'm not biased towards it. Never shot with it. (Though I would have if UHD were possible.)

 

[NorBro]

ML is just for fun, on your own time. B-roll, personal projects, few seconds here and there. You wouldn't want to use it for anything serious if worried about Blackmagic's reliability.

My one takeaway from that channel is everything looks the same, lol. In a good way...whether it's from a $500 camera or a $5000 one (grading is a part of that).

On a side note, I like the FX9 more and more each week. It has a look, there is something to it (more than just grading).

Footage I caught today: https://www.youtube.com/watch?v=2WLdXxgexy0

Wonder how these upcoming small Sony powerhouses will look.

 

[ahalpert]

Nobody is saying ML RAW is a reliable workflow. I just like the way it looks.

Sony has come a long way in terms of the look. Cinetone looks really nice to me and seems to lend itself to a more mellow, cool palette than Canon. That clip looks nice - but that's not cinetone, that's something else right?

Sony seems to be having trouble with high framerates. Can't even do no-crop 60p from the FX9. I wonder what the Sony DSLMs will do but the R5 outperforms the FX9 on resolution and framerates (though not necessarily on IQ). My expectations are tempered. But it would be very convenient for me if the A7SII successor were preferable or on par to the R5.

 

[NorBro]

S-Log3.

R5's paper specs...again...don't mean much. Canon's C500 Mark II is $16,000 with the same boring color science (but nicer resolution) we've seen for the last 10 years (IMO).

Nothing exciting about it (although that is a very good thing for some people).

FX9 is getting a full-frame 60p update in October.

 

[ahalpert]

S-log3 but conformed to a LUT other than Cinetone

Specs mean specs. It ain't shooting 720p 30fps. C500mk2 footage looks pretty nice to me FWIW.

FX9 can only shoot 4k 60p with a 1.2 crop after firmware update. People need to stop calling it full-frame. It's going to be a real problem when switching between regular and slow-mo, especially in ENG/run-and-gun. It's not cool for a camera in its class. Maybe if it was a 1.05 or 1.07 crop, but it's not. For some reason, Canon has started trouncing Sony with sensor readout or data throughput or something.

 

[NorBro]

Regarding the FX9, it's close enough, lol - but I do agree, crops should not exist in its class and any should be referred to correctly for the highest level of professionals.

As far as 'specs mean specs'...I think that is a contradicting note based on all of your thoughts and questions here.

 

[ahalpert]

It's not close enough. A 20 becomes a 24. Aesthetically it's close enough, but it screws with your frame size and will rob you of precious FOV on the wide end.

Specs mean specs...if you need to deliver 4K and your camera doesn't, you're SOL. If you want to shoot 120 fps in 4K and your camera doesn't, you're SOL. If you are relying on AF and it disappears at higher framerates, you're SOL. Specs don't always correlate with IQ, but they're far from meaningless.

Just look at the 4K 30p 8-bit codecs on the current Sony "powerhouses" or the 4K crop factors on Canon DSLR/mirrorless. No fun!

 

[NorBro]

It's close enough for most people using that camera. When you're in Hollywood you can worry about the differences if you're using a FX9.

Do you think $50,000 REDs shoot all of their high-framerates with the entire sensor?

And I think reliable specs mean reliable specs.

Because specs certainly do not mean specs unless you're willing to lie to people or only pass on information from a company and what it's telling you your camera has...or you're just dumb (not you).

 

[ahalpert]

I disagree re:FX9 users. I think you probably have more flexibility in a narrative context to swap lenses or have lens options on hand to deal with weird croppings. FX9 seems more like doc/eng/corporate camera. Need to move fast.

Sometimes the RED crops cause problems, but they generally have pretty high FPS using full sensor. The higher FPS crops are a bonus.

Maybe the R5 will have unusably bad 60p or 120p from the full sensor and in that sense the specs are a lie. But the FX9 has nothing.

And yeah the 1DXmk3 is an odd duck with borderline unusable rolling shutter in FF up to 30p and then better performance at higher framerates. Always little surprises. But my S1 must crop to APS-C to get 4K 60fps. Life is complicated.

 

[AndreeOnline]

Can anyone tell me why canon 5D magic Lantern RAW footage has so much richer color than anything else?

OK, Abe...

I became curious myself to see where the rubber meets the road. I want you to look at this short video.

BUT: please understand that these are not "Goldilocks grades" that are 'just right'. Especially a few of them, as scene brightness gets overall darker, these grades are overbearing and saturated. Each clip is preceded by an ungraded version that serves as reference for when grades get wonky and annoying.

The question, then, is: are these colors "thin"? Is there a difference between rich colors and saturated colors?

The ProRes 422HQ ended up being 8GB in 4K DCI, so I couldn't upload that (capped at 5GB). You can however download a 4K 10bit HEVC. I have no idea how Vimeo will treat this in the conversion, so the downloaded file is recommended.

https://vimeo.com/429611989

 

[ahalpert]

Thank you, Andree, that was very thoughtful. The P6K always looks great and makes me want one. That night footage, especially - oversaturation aside - looks fantastic.

It's possible that I'm deluding myself and there's no there there. But I think that what I may be responding to in the ML RAW footage and describing as "rich" is tonal variation.

To test that, I compared some of these clips in vectorscope. Not scientific - no matching of the scene or the grade. But my thinking was that an image with more tonal variation would have a more dispersed data plot in the vectorscope.

Here's a P6K shot from you:


I think the vectorscope looks more concentrated than some of these ML RAW shots:





Here's a P6K shot from you that had a dispersed vectorscope:


But that might only because of the different objects in the scene and not color variation within individual objects?

Here's looking at a slice of clouds from that shot:


Anyways, it's just a theory. I could be full of crap.

Hmm, hard to evaluate those vectorscopes at this resolution. Here are the screenshots in google drive: https://drive.google.com/drive/folders/1B6SGgeusYLmiVHSZcD8i0EHj3AKe5889?usp=sharing

 

[AndreeOnline]

I think the vectorscope looks more concentrated than some of these ML RAW shots

In sunset type scenarios I see similar data:

No adjustment here, apart from boosting saturation in the raw tab.

 

[ahalpert]

Maybe. But I feel like the ML RAW data more resembles a cone or a cloud to me. But at this point, I feel like I'm too invested in proving my point to be an impartial judge. That sunset shot you posted looks really nice. Does it look AS nice to me as the other sunset colors? I'm not sure.

 

[ahalpert]

OK Looking at it more closely on my laptop and not my phone...there is definitely some fuzz at the edges of the color distribution on the P6K, which is good - that indicates variation. But I do believe that the P6K is more concentrated in its hue distribution in one part of the vectorscope, and in its saturation distribution in another. It is less diffuse.

P6K:


ML Raw: