C500: 1D-C Users

  • Thread starter Thread starter Guest
  • Start date Start date
MattDavis said:
Hmm, that sounds familiar - like the 1D-C? Where some say one needs to overexpose in C-Log to get a better looking 'digital negative'?
Click to expand...

Doing shadow tests, c-log has more noise, and less detail than neutral in 4k mjpeg at ISO 400.
 
Last edited:
This image is taken from underexposed frames of 4k mjpeg footage. They do not represent the image from the camera, just the difference between the two settings of Neutral and C-Log. The images were altered to show shadow detail and match contrast.


ShadowVS_Neutral_4k by onesielaff, on Flickr

What didn't surprise me, was that the raw still had more shadow detail than C-Log. What did surprise me was the difference in shadow detail between C-Log and Neutral.

Neutral has better detail, gradation, and noise performance than C-Log at iso400 in the shadows. So, C-log magically has more highlights than raw, but less shadow detail than Neutral picture style. It would suggest that C-Log is underexposing. Hope I am not spreading misleading information here. Just my experience.

Now the main thing is "degree"... to what degree is the shadow versus highlights going to win out.

(Due to C-Log needing to be pushed farther to get these results, it will obviously look worse than Neutral. The main idea was to look for 'recoverable' information.)

Here are the original exposures:


With the other codecs, recovering details in the shadows seemed less worth while, as the blocks were bigger (that was in part due to resolution of 1080p versus 4k as well). PSE is showing more shadow info recovery from a raw than C-Log, as well.
 
Last edited:
Initially, I was under the assumption that Canon would not mess with exposure between the settings. Mostly, i was hoping all that highlight recovery was free. Since I often found myself underexposing to recover highlights, I was just looking at the highlight recovery between the settings. But after finding that C-Log was getting more highlight info than raw files, I was interested to see what was happening in the shadows. AndreeOnline expressed similar thoughts, and explained them in good detail. As it turns out, C-Log is not recovering as much as other settings are in shadow detail, suggesting that C-log is underexposing compared to the other settings. So, now to test the difference between different exposures and see how things go.
 
Last edited:
I'm not sure C-Log is better than the profiles. I think the reason to use it would be not so much if you wanted more range, but if you wanted something distinct from every one of the profiles. It might be better in that case to shoot C-Log and effectively create your own profile in grading. But if you might be nearer to one of the existing profiles, I think it could be better to shoot in that profile, sort of like doing it 'in camera' as in shooting on film or whatever. I just shot a feature entirely in one of the profiles.

The reason I feel this is basically their use of a codec at this resolution. I personally think they have used the codec very nicely. But it's still a codec and not raw. It's 8 bit too! Onto memory cards. So there's a certain amount of ingenuity involved, a bit of a trick taking place. And I feel having C-Log on the 1DC despite this is Canon kind of saying... "well, you try then!" and I think that is very mature of them. I mean, it's not hard to imagine this very camera being on the market, at this very price, but, eh, no C-Log...

I'm glad it's there and I may shoot a whole movie with it if I can get better results, but at the moment I find that some of the profiles are actually like shooting on Kodak! It's really a lovely image this camera has. That was something that was accepted with film cameras and stocks, that they each had a distinctive look but I would have laughed if someone tried to get me to accept it (in a positive sense) about video/digital video to date.

Not anymore!
 
edfwfwefwef said:
I'm not sure C-Log is better than the profiles. I think the reason to use it would be not so much if you wanted more range, but if you wanted something distinct from every one of the profiles. It might be better in that case to shoot C-Log and effectively create your own profile in grading. But if you might be nearer to one of the existing profiles, I think it could be better to shoot in that profile, sort of like doing it 'in camera' as in shooting on film or whatever. I just shot a feature entirely in one of the profiles.

The reason I feel this is basically their use of a codec at this resolution. I personally think they have used the codec very nicely. But it's still a codec and not raw. It's 8 bit too! Onto memory cards. So there's a certain amount of ingenuity involved, a bit of a trick taking place. And I feel having C-Log on the 1DC despite this is Canon kind of saying... "well, you try then!" and I think that is very mature of them. I mean, it's not hard to imagine this very camera being on the market, at this very price, but, eh, no C-Log...

I'm glad it's there and I may shoot a whole movie with it if I can get better results, but at the moment I find that some of the profiles are actually like shooting on Kodak! It's really a lovely image this camera has. That was something that was accepted with film cameras and stocks, that they each had a distinctive look but I would have laughed if someone tried to get me to accept it (in a positive sense) about video/digital video to date.

Not anymore!
Click to expand...

I have to agree with you. This is one of my favorite cameras.
 
If anyone is doing more standard (Charts, standard light kits, etc) tests on the dynamic range or latitude of C-Log vs other picture styles, feel free to chime in or correct me at anytime. My tests are much more geared to my personal needs and so they are a bit lacking in hard numbers and measurable differences.

From comparing C-Log and Neutral at different exposures, it does appear there is almost a 2stop difference in their exposure levels. When Neutral is exposed 2 stops under C-log, the difference between highligh retention and shadow in the images is reduced. I still give the edge to C-Log.

The comparisons in 4K Mjpeg is where it is at, as the Mjpeg codec and resolution of 4k seem to maintain much more detail and allow for more post manipulation than either IPB or ALL-I. So pushing the shadows in post with IPB will not get you as far, as artifacts appear earlier. But the the s35 image is still on another level from all previous dslr's, as has been stated many times.

When I pushed both image settings to get maximum DR, I found C-Log rendered more detail with less noise. But picture styles with shadows lifted like flaat2 may even that out a little, as flaat2 digs into the shadows before the codec. Either image setting displays banding on gradients when the curve does not suite that particular gradient. But I have to get more real world tests in, as I doubt I would ever push footage that far.


While C-Log may be ~trickery~, or not, I like the potential theory behind it. But all previous tests showing those highlights gaining almost two stops are not quite so representative of reality, imho.
 
Since I tend to underexpose, when I switched to C-Log I was having issues with the post image. Turns out clog was already underexposing for me, and that made for some really underexposed footage, which had to be extremely manipulated in post to bring back. (one example is the shot of the Chinese man smoking)
 
James0b57 said:
Since I tend to underexpose, when I switched to C-Log I was having issues with the post image. Turns out clog was already underexposing for me, and that made for some really underexposed footage, which had to be extremely manipulated in post to bring back. (one example is the shot of the Chinese man smoking)
Click to expand...

True! when C-log is enabled, 18% gray will be automatcially remapped from 50% IRE to about 32%. Expose normally with VA on and you will have a spot on C-log.

Shane Hulburt says it's better to slightly overexpose with C-log on 1D C.
 
marvinhello said:
Shane Hulburt says it's better to slightly overexpose with C-log on 1D C.
Click to expand...
I underestimated what that meant, Haha!

A few guys I know over expose on digital cameras, so they may actually do better with C-Log right out of the gate. I have become accustomed to under exposure. But may adjust for c-log.
 
It is fun playing between C-log and PS. They are different exposures for sure. Roughly 1.5 stops different in practice. I find when I expose a full 2 stops over (whether by changing aperture or iso) typically I lose highlight detail and gain shadow detail in one or visa versa.


Since many people I shoot have beautiful black hair, I tend to value that shadow information when I have the option to keep it. So, I have to balance DoF and ISO when composing some shots. Theoretically raising the iso would reduce DR in C-Log...

Here is a shot trying to break the 1D-C, but it performs admirably. Trying to balance detail in the hair with a tasteful amount of blowout from the window:

Zeiss 50mm f4/5.6 split, no filters
ISO400, Neutral (0,-2,-2,0), 1/50. 24fps 4kmjpeg
WB daylight preset, no WB shift, no tone shift
No CC, JPEG still capture from timeline...

of course this would be difficult for any camera, and if this were not a natural light shoot, I could easily use some fill or accent, if I needed more detail in the hair.
 
Last edited:

Shadow detail. Fine details in hair. Looks like iso800 would've been best choice for C-Log.

At iso 400, Neutral is properly exposed, while C-Log is underexposed, losing vital information in the shadows, but bringing in amazing detail in the highlights (naturally).



C-Log reigns in those highlights. Creamy.

Overall, I feel C-log does give more for post work, but the end results are similar enough to go either way. The interesting thing is how they differ in exposure, allowing for more creative choices of DoF. Though, noise performance is good, I am going to explore the value of shooting in Neutral at iso1600 versus C-log at iso3200 or more.

C-log at iso6400 is usable, but not desirable, imho. But, I would say the same thing about many other cameras at iso1600. 1DC is great.
 
Last edited:
James0b57 said:

Shadow detail. Fine details in hair. Looks like iso800 would've been best choice for C-Log.



C-Log reigns in those highlights. Creamy.
Click to expand...

Could you do a comparison between C-log with LUT (official Canon one or FilmConvert Pro) and Neutral with exact exposure setting?
 
marvinhello said:
Could you do a comparison between C-log with LUT (official Canon one or FilmConvert Pro) and Neutral with exact exposure setting?
Click to expand...

Here is the C-Log versions of the same shot with Filmconvert C300 C-Log LUT applied:

Same settings as same shot in Neutral, except for C-Log applied:

Zeiss 50mm, f4/5.6 split no filters
iso400, WB daylight preset, 24fps, 4k Mjpeg


(C-Log iso400, Filmconvert C300 C-Log default LUT applied)





and same shot, except in Neutral with no LUT applied:


(Neutral, no changes in post, in-camera look)




And here is same shot, with C-log at iso1600:

(C-Log iso1600, Filmconvert C300 C-Log default LUT applied)

unfortunately, I did not do any exposures at iso800 with C-Log, as that would have been approximately the same exposure as Neutral.
 
Last edited:
MattDavis said:
Hmm, that sounds familiar - like the 1D-C? Where some say one needs to overexpose in C-Log to get a better looking 'digital negative'?

Apropos of all these contra-indications, 'Negative Gain' in Sony cameras saw clipping in the shadows, and I (wrongly) assumed the same would be for Canon. But the charts and collateral didn't seem to make sense. I could only 'square the circle' if this was not the electronic equivalent of 'push processing' but, in fact, 'pull processing'. Wouldn't make sense in chemical terms, but if you're just mapping photon accumulation on a sensor to points on a gamma curve? So 'negative gain' (or lower-than-standard ISO) just a LUT cheat?
Click to expand...

Shooting Sony (PMW-350) a little while back, the camera came to me setup with the REC709 gamma setting. In some ways very nice and punchy, but far and away too clipped and crispy. The shadows were mascara black and the highlights turned yellow before falling off a cliff to their (your) death, never to be seen or heard from again.

I quickly turned to looking into the Hyper Gammas, and found the grey point very different (compared to the standard gammas). There was the initial excitement of seeing amazing highlight retention then, I realized my mids were all under. Meaning, in order for me to get a similar exposure in the mids, and get true black back, I had to overexpose a bit, stealing from the highlights range I had just recovered. Seeing as I could not do much post work, (news), I had to come up with a different solution. So, I ended up with something much more moderate, stealing many ideas from XDcamuser. In some low light, low contrast situations went back to rec709 (shudder~).

Log and hyper gammas, are pretty different animal from raw or other picture styles, and require some familiarity. Even though, it may just be an exposure adjustment to some, infact the footage behaves differently. It is like starting to work with an entirely different camera. Raw is the easiest in my opinion, since it is a fairly linear representation of the sensor. I recently worked with a guy that liked S-log, but found C-log difficult to finesse, citing it was too saturated. Seeing as C-log is newer, I assume it has a lot to do with how much time he has spent with it, and potentially 8bit vs 10bit design.

Personally, I dislike log (especially in 8bit), but if I can get an extra 1/2 stop or more, then I'll take it.
 
Am I the only one getting actual underexposure from C-Log? I see no one else mention it. Thanks
 
You must log in or register to reply here.
Back
Top