Welcome back to “In-Depth with the C500 MKII” blog post series. In this post, I will be covering footage analysis. If you haven’t checked out the rest of the series, I highly recommend starting with Part 1, where I outline all the topics covered from beginning to end.
Post 3 - Footage Analysis:
Scaling Sharpness
Since the C500 MKII has so many options to choose from in terms of Codec (RAW Light and XF-AVC), Sensor Size (Full Frame, Super35 and Super16) and Resolution (DCI 4K, UHD, 2K and HD), I found it important to compare how these various formats handle when paired with another in a particular resolution. I see myself constantly switching formats depending on the shooting situation. Maybe I’ll be shooting an interview in 5.9K Cinema RAW Light Full Frame, vérité in XF-AVC Super35mm in 4K and slow motion over 60fps in S16 Crop in 2K. The question is, how will these formats intercut and is there a noticeable sharpness difference between scaling RAW or XF-AVC.
I have selected 4 comparisons to look at, given the way I normally shoot and the deliverables most used these days. The first option is a full 5.9K sequence, scaling the XF-AVF Full frame 4K and 2K accordingly. The second option is a DCI 4K sequence using Full Frame files; scaling the 5.9K RAW and XF-AVC Full Frame 2K accordingly. The third option is a DCI 4K sequence using S35 Crop files; scaling the S35 2K accordingly. The final option is a DCI 4K sequence using S16 Crop files; scaling the RAW S16 Crop, XF-AVC S16 Crop 2K accordingly. Most, if not all of the work I do these days finishes in 4K, so I did not include a 2K scaling option here. If you find yourself finishing in 2K/HD most often, I urge you to do your own tests to see how they handle.
Let’s take a look at how these formats compare to each other, depending on the finishing format you are going for.
5.9K Sequence
*All images enlarged 400%
DCI 4K sequence using Full Frame files
*All images enlarged 400%
DCI 4K sequence using S35 Crop files
*All images enlarged 400%
DCI 4K sequence using S16 Crop files
*All images enlarged 400%
In all scenarios, the RAW files appear to be sharper than the XF-AVC files, even at the same resolution. If you take both the S35 Crop and S16 Crop examples, you can see more definition in the bricks of the RAW than you can in the XF-AVC. It isn’t a huge difference, and one that could probably be adjusted for with post sharpening, but it is a difference that you can see.
What’s interesting is the 4K Full Frame sequence. Since the camera is oversampling the 5.9K sensor when shooting XF-AVC Full Frame 4K, the sharpness difference appears to be on par with the straight 4K to 4K comparison I previously mentioned, except here the 5.9K file is scaled down to 69%. I’m impressed how well the XF-AVC holds up in this comparison. This is great to see, as I will probably be using XF-AVC more than RAW for documentary production to save hard drive space.
Dynamic Range
Canon has always touted their Log 2 gamma curve to have 15 stops of dynamic range, where Log 2 is supposed to come in around 14 stops. In my previous test of these curves with the C300 MKII, I did not necessarily find that to be the case, especially in a real world environment. I decided to conduct my test with the C500 MKII in a similar fashion.
Something interesting to note is that in Cinema RAW Light the camera does not shoot below ISO 800. If you are shooting in XF-AVC, you can take the ISO all the way down to 160. Something to keep in mind when shooting RAW.
For the dynamic range test, I set up a gray card and lit it evenly with a soft light source with an exposure of T16 at ISO 800. I set the camera to Canon Log 2 in Cinema Gamut and shot with a Canon Sumire 85mm. I then over and under exposed the camera one stop at a time until I flat-lined at either 0% or 100% on the IRE scale. There will probably be some minor fluctuations in values, depending on the lens quality and f-stop characteristics, but these are factors that will exist in the real world, so I am fine having them part of my test. Because remember, the real reason for testing all of this is to give you a more accurate understanding of the equipment you are working with.
All of the charts below reflect the Log readout of the camera with no LUTs applied.
RAW Full Frame – Log 2 / Cinema Gamut
In RAW Full Frame at Log 2 / Cinema Gamut, I am seeing 15 distinct exposure changes from 6 stops over-exposed to 8 stops under-exposed. The last stop under is a minimal shift, but it is perceivable. This is showing me a dynamic range of 15 stops.
RAW Full Frame – Log 3 / Cinema Gamut
In RAW Full Frame at Log 3 / Cinema Gamut, I am seeing just about 14 distinct exposure changes from 6 stops over-exposed to 6 or 7 stops under-exposed. The jump from 5 to 6 stops under-exposed is noticeable, but the increment between 6 and 7 is ever so slight. I’m putting this around 13.5-14 stops of dynamic range.
XF-AVC Full Frame 4K – Log 2 / Cinema Gamut
Moving into XF-AVC Full Frame DCI 4K at Log 2 / Cinema Gamut, I am seeing 15 distinct exposure changes from 6 stops over-exposed to 8 stops under-exposed. Here, the shift between 7 and 8 stops under-exposed is more noticeable than is was in RAW. This again puts us at a dynamic range of 15 stops.
XF-AVC Full Frame 4K- Log 3 / Cinema Gamut
In XF-AVC Full Frame DCI 4K at Log 3 / Cinema Gamut, we still have 6 distinct stops of over exposure, but that subtle difference between between 6 and 7 under-exposure is gone. I’m putting this at a solid 13 stops of dynamic range.
It’s interesting to see how the log curves handle differently in RAW versus XF-AVC. I would of course assume that RAW would give more latitude, but I’m surprised to see Log 2 handled exactly the same. It’s only Log 3 that loses a half to a full stop of range in the shadows. This is all good news in my book.
Exposure Adjustment
The next thing I wanted to look at is how the camera handles correcting for over and under-exposed shots. While the dynamic range sits high, does the camera correct well for shots not perfectly exposed?
I used Davinci Resolve to perform these tests. With Cinema RAW Light files, I made the exposure correction by raising or lowering the exposure adjustment in the Camera RAW panel. You could of course tweak this even more with Lift, Gamma and Gain adjustments, but I wanted to see where I would land with just utilizing the RAW exposure method.
With XF-AVC files, I made the exposure correction by adjusting Lift, Gamma and Gain until it matched as closely as possible to the At Exposure shot.
I balanced each shot while in Log, and then added Canon’s Log 2 / 3 Rec709 LUT to normalize.
Cinema RAW Light
This first set of shots is correcting Log 2 footage over and under-exposed back to match the At Exposure clip:
What is amazing about this test is how far I can correct an over-exposed image. At 5 stops over, I am able to normalize it with little to no impact in the image. I am starting to lose information in the highlights, but the rest of the shot looks great. It’s not until correcting the shot 6 stops over that things start to fall apart. The opposite is true for correcting under-exposed shots. Correcting something a stop under-exposed is no problem, but with 2 stops under you start introducing noise. By 3 stops under, the image becomes incredibly noisy.
Now let’s look at shots in Log 3 over and under-exposed back to match the At Exposure clip:
The Log 3 footage acts the same as Log 2, in that it can handle over-exposed images much better than it can under-exposed. The difference with Log 3 is that you have about 1 less stop of play on either end. The shot 4 stops over-exposed looks great normalized, but it starts to fall apart 5 stops over. When under-exposing, you can only get away with about a stop before the introduction of noise becomes a serious issue.
XF-AVC
Let’s look at the same test with XF-AVC footage, starting with Log 2:
First off, I want to remind everyone I am not a colorist. But I do hope with that in mind, these examples give you a good enough idea of how the camera handles in these conditions.
In my adjustments, it seems that the XF-AVC Log 2 footage handles similar to the RAW, except that you have about 1 less stop to work with when over-exposed. I am able to draw a decent normalization at 3 stops over, but by the time I get to 4 stops over, there isn’t much left to work with. When it comes to normalizing the under-exposed images, it seems to work exactly the same as the RAW. 1 stop under is fine, 2 under and noise is introduced (although a less than the RAW), and by 3 stops under, there’s a ton of noise.
Now let’s look at shots in Log 3 over and under exposed back to match the At Exposure clip:
The XV-AVC Log 3 footage handles the same as the Log 2 footage did in comparison to the RAW. We have about 1 less stop to play with in the over-exposed frames compared to XF-AVC Log 2, and about 1 less stop compared to the RAW in Log 3. In the under-exposed frames, the noise actually holds up better at 3 stops under than it has in any other example.
Thoughts
My overall thoughts with how the C500 MKII handles over and under-exposure is that across the board, it is much much safer to over-expose your images than under-expose. This is no different than with the C300 MKII. Based on these tests, I would recommend even pulling “the stock,” by shooting a stop over-exposed and adjusting the exposure in post. You will need to create a LUT that shows you a “properly exposed” image when doing so. By purposely over-exposing your image, you will reduce noise across the image, as well as protect your shadows from excessive noise, even when they are properly exposed.
When it comes to the question of when to shoot in Log 2 versus Log 3, I would say without a doubt shoot in Log 2 to gain the most latitude of your image. Now that the C500 MKII doesn’t even let you shoot lower than ISO 800 in RAW, I see no reason not to use Log 2 all of the time when you need the most out of your image.
Where this answer really gets answered (or at least balanced against) is how well the camera handles noise at various ISOs. Let’s take a look at that next.
Noise
On the C300 MKII, the main reason I would ever shoot Log 3 is if I am in dark situations. In Log 2, noise was introduced above ISO 400 which made it difficult, if not impossible, to use indoors. With Log 3, I could comfortably push it to ISO 3200, and in some extreme situations would even go as high as ISO 8000.
With the C500 MKII, I’m curious to see how the noise levels hold up, especially since the camera is limited to ISO 800 and above in Cinema RAW Light.
Each section will have footage in its native log gamma, as well as an export with Canon’s 709 correction LUT applied. You will notice that the noise appears more prevalent in the Log files, but once you move to Rec709, a lot of that is mitigated. All of my notes below are talking from the point of view of Rec709, but the Log versions are there for your reference.
**A note – there seems to be a strange conversion to RAW footage shot in Log 3 when using the 709 LUT. Log 2 footage processes fine, but Log 3 seems to be lifted. This does not occur when using the same LUT on XF-AVC Log 3 footage**
RAW Full Frame – Log 2 / Cinema Gamut
Unlike the C300 MKII, ISO 800 is very clean. Noise begins to get introduced at ISO 1600 and ISO 2500 seems to be the comfortable limit to push before the noise becomes unruly. You could probably get away with pushing to ISO 3200-4000 if you crush the blacks and do some noise reduction in post. This is definitely an upgrade from the C300 MKII.
RAW Full Frame – Log 3 / Cinema Gamut
Log 3 in RAW seems to not handle well at all. ISO 800 is clean, but noise becomes to get introduced around ISO 1000. ISO 2500 seems to be the comfortable limit to push before the noise becomes a problem. But here, I would only go to ISO 3200 if you crush the blacks and do some noise reduction in post. This is definitely a downgrade from the C300 MKII.
XF-AVC Full Frame 4K – Log 2 / Cinema Gamut
Oddly enough, Log 2 in XF-AVC handles exactly the same as in RAW. ISO 800 is clean before noise gets introduced at ISO 1600. ISO 2500 seems to be the comfortable limit to push before the noise becomes unruly and would only go as high as ISO 3200-4000 if you crush the blacks and do some noise reduction in post.
XF-AVC Full Frame 4K – Log 3 / Cinema Gamut
Log 3 also handles the same in XF-AVC as in RAW. ISO 800 is clean, but noise becomes to get introduced around ISO 1000. ISO 2500 seems to be the comfortable limit to push before the noise becomes a problem. But here, I would only go to ISO 3200 if you crush the blacks and do some noise reduction in post.
After seeing these results, I 100% stand by my suggestion of always shooting Log 2, regardless of whether you’re in RAW or XF-AVC.
Purple Banding Solved!
As many of you are aware, the C300 MKII had a debilitating issue of purple streaks appearing in the frame when shooting a subject against a strongly lit background. I found this problem to occur whenever I shot someone driving a car. Even with ISO changes, exposure changes, or Log changes, I was never able to make this go away. It was important for me to test and see if this problem occurred with the C500 MKII.
Below is a screen grab of the C300 MKII in Log2 / C.Gamut with no correction. You can clearly see the purple band over the lower left portion of the frame. This is very difficult, if not impossible, to remove in post and despite a 709 conversion crushing the blacks, the purple streak is not eliminated.
Here are screen grabs from the C500 MKII in Log2 / C.Gamut, in both RAW Full Frame and XF-AVC Full Frame, shot on the same stretch of road to replicate consistent shooting conditions.
I am pleased to report that I did not find the issue to occur with the C500 MKII. The purple band would appear across the entire shot with the C300 MKII, so its not a frame-by-frame anomaly. With the C500 MKII, I never came across the problem in any format, in any frame, throughout the entire shot. This should be a huge sigh of relief for anyone who has struggled with this issue in the past.
Conclusion
The C500 MKII has really gotten an upgrade in terms of its resolution options, Full Frame mode and ability to shoot in Cinema RAW Light. The ability to mix and match these formats in a single project is nice and RAW really shines when it comes to resolution and sharpness. Of course the biggest drawback is the size of the files.
It also seems that Log 2 on this camera has really gotten an upgrade. I can now safely say the claim of 15 stops of dynamic range is valid and it gives you a ton of room to correct for over-exposed images. Sadly though, it still underperforms in the darker areas of the frame. Log 2 is also much better in higher ISOs than it previously
was. Being able to push higher allows for even more exposure protection in the shadows.
Log 3, on the other hand, leaves a lot to be desired. I was such a big fan of it in the C300 MKII for allowing me to shoot at high ISOs, but that doesn’t seem to be the case anymore. And while it has somewhere between 13.5 to 14 stops of dynamic range, it underperforms in over and under-exposure compensation compared to Log 2. Also, the strange interpretation of RAW Log 3 footage when using the 709 correction LUT leaves thinking I won’t be using this at all.
It’s also fantastic seeing that the purple banding issue that was so crippling on the C300 MKII has been resolved with this sensor.
All in all, I think its safe to say Log 2 is the way to go and XF-AVC is an awesome compromise to RAW when you need to save space on the hard drive while still getting an incredible image.
As always, I encourage you to do your own tests and not just take my word for it. I am curious to see if you come to the same conclusions. Please leave me a comment with any interesting findings you might come across!