Canon C70 User Experience

[herein2020]

50 minutes ago, Mmmbeats said:

I did some proper low light shooting with the camera (a completely unlit tunnel deep underground, lit only by somebody's helmet torch). The results blew me away. I truly think c log 3 is redundant on this generation of cameras. 

You are probably right, the differences in CLOG2 vs 3 on this camera are probably so negligible that it would be easier to just stick with CLOG2 even if on paper CLOG3 would technically be the better choice. My whole workflow is around CLOG2 so I probably will just use it for everything.

 

52 minutes ago, Mmmbeats said:

Regards the straight adaptor, the simplest one (no drops-ins, etc.) is only £80. Probably worth getting just for the extra flexibility it offers. 

 

I only have a single APS-C lens that would use it, and I would have to unbolt the speedbooster so I will probably skip it for now. If I do end up replacing my 5DIV with an R5 or R3 then I would pick up the straight through adapter.

[BenEricson]

2 hours ago, herein2020 said:

You are probably right, the differences in CLOG2 vs 3 on this camera are probably so negligible that it would be easier to just stick with CLOG2 even if on paper CLOG3 would technically be the better choice. My whole workflow is around CLOG2 so I probably will just use it for everything.

I find that the Clog 3 / BT.709 profile is easier to color grade from scratch. That being said, I think Log2 is a better curve.

I really liked log2 bt.709 tho. It looks a bit closer to Red or Arri Log footage. This was only available on the C300 Mkii. Maybe the c500ii? Not sure.

[herein2020]

2 hours ago, BenEricson said:

I find that the Clog 3 / BT.709 profile is easier to color grade from scratch. That being said, I think Log2 is a better curve.

I really liked log2 bt.709 tho. It looks a bit closer to Red or Arri Log footage. This was only available on the C300 Mkii. Maybe the c500ii? Not sure.

 

I found VLOG very easy to grade from scratch, but with CLOG2, the Buttery LUT pack was a better fit for me, I did a few test hand grades and didn't like the results. The Buttery LUT gets me a good CLOG to Rec709 grade then I finish the look with a creative grade.

[M_Williams]

14 hours ago, Django said:

it is until it hits the speed booster. a speed booster adds an extra glass element changing the optical design of the original lens. It demagnifies and hence concentrates light adding a stop when it hits the sensor. this article explains it better using the same 50mm example too:

I've already explained above that the entrance pupil does not change. However, given that we've now reduced the focal length (from 50 to 35), the formula for calculating f-number has changed. The entrance pupil is still 25mm in diameter, so it now gives us 35/25=1.4. So a 50mm f/2 lens, speedboosted, becomes (not effectively, but actually) a 35mm f/1.4. That's how Speedboosters 'magically' create an extra stop of light. To clarify, our 35mm f/1.4 will give you an exposure one stop brighter than the original 50mm f/2, whether that 50mm is used on full frame or S35.

https://www.alex-stone.com/2020/10/29/busting-speedbooster-myths/

 

That explanation is.... literally what I said. 50 -> 35, and f/1.4 (in the case we're talking about) becomes f/1.0. So, effectively, a 35/1.0. That's precisely what I said, and not what the OP said (he said 50/1.0).

Also, it is indeed still transmitting the same TOTAL light. Not light per area, but TOTAL light. The APS-C sensor in the camera (with a speedbooster) now collects as much light as a FF camera without the speedbooster. That's... how speedboosters work. They're essentially the opposite of a telecoverter.

[M_Williams]

13 hours ago, herein2020 said:

 

You are saying the same thing I said...all I said was the lenses gain a stop of light but without gaining the typical shallower DOF, regardless of how the math works out at the end of the day the 50mm gains a stop of light and is still a 50mm FOV without a shallower DOF at F1.0.  And yes, the 50mm is 1 stop brighter (F1.0) than if it were mounted to a camera without the speedbooster.

I refer back to the same video that I already posted that shows the exact same thing I said:

 

 

 

I have read that in a true lowlight situation that CLOG3 would be the better color profile, but I haven't shot at night yet with it so I have no idea. 

Yes, they gain a stop of light, but they ALSO give a wider FOV. That wider FOV reduces the DOF from the same distance-to-subject. So it's now a 35mm FOV (on FF) with an f/1.0 aperture, which is exactly the same DOF as a 50mm f/1.4 on FF (from the same distance).

A speedbooster DOES give shallower DOF. But it can't give you shallower DOF (or a wider FOV) than the lens is designed to render on a full-frame sensor.

I just don't understand the part of "without the typical shallower DOF." Like... a speedbooster DOES do that. It matches exactly what it would be on a larger sensor - it can't do more than that. 35/1.0 = 50/1.4 for DOF at the same distance.

[herein2020]

56 minutes ago, M_Williams said:

Yes, they gain a stop of light, but they ALSO give a wider FOV. That wider FOV reduces the DOF from the same distance-to-subject. So it's now a 35mm FOV (on FF) with an f/1.0 aperture, which is exactly the same DOF as a 50mm f/1.4 on FF (from the same distance).

A speedbooster DOES give shallower DOF. But it can't give you shallower DOF (or a wider FOV) than the lens is designed to render on a full-frame sensor.

I just don't understand the part of "without the typical shallower DOF." Like... a speedbooster DOES do that. It matches exactly what it would be on a larger sensor - it can't do more than that. 35/1.0 = 50/1.4 for DOF at the same distance.

 

I still think you just don't get what I said but it's ok. You are focusing on the conversion process and I simply skipped to the end and was focusing on the end result. The end result is that you have a 50mm FOV with an F1.0 light gathering ability but with the 50mm equivalent DOF of F1.4 vs F1.0 when mounted to the C70.

The fact that the lens is converted to a 35mm lens wouldn't be relevant to me unless it was mounted to a FF sensor because the effective FOV on the C70 which is the camera being discussed is 50mm.

[M_Williams]

13 minutes ago, herein2020 said:

 

I still think you just don't get what I said but it's ok. You are focusing on the conversion process and I simply skipped to the end and was focusing on the end result. The end result is that you have a 50mm FOV with an F1.0 light gathering ability but with the 50mm equivalent DOF of F1.4 vs F1.0 when mounted to the C70.

The fact that the lens is converted to a 35mm lens wouldn't be relevant to me unless it was mounted to a FF sensor because the effective FOV on the C70 which is the camera being discussed is 50mm.

Then I just don't understand the problem here. It's like you want something that isn't physically possible. "with the 50mm equivalent DOF of F1.4 vs F1.0"... yes, because it's really a 35mm f/1.0. It's like you want the extra light but somehow also shallower DOF than you'd get on FF. That's not possible, because the two are inherently linked.

Like, the end result is exactly what you'd expect. You ARE getting the DOF of a F/1.0 lens. But not a 50/1.0.

And you seem to be missing the point of it being 35mm. A 35mm f/1.0 on FF has the same DOF (from the same position) as a 50/1.4 on full frame. With it being on the C70, it returns to a 50/1.4 (in DOF terms), which is exactly what the lens is. The speedbooster can't give you anything more than the lens is capable of natively.

[herein2020]

1 minute ago, M_Williams said:

Then I just don't understand the problem here. It's like you want something that isn't physically possible. "with the 50mm equivalent DOF of F1.4 vs F1.0"... yes, because it's really a 35mm f/1.0. It's like you want the extra light but somehow also shallower DOF than you'd get on FF. That's not possible, because the two are inherently linked.

Like, the end result is exactly what you'd expect. You ARE getting the DOF of a F/1.0 lens. But not a 50/1.0.

No it really seems like you just didn't understand my post at all....I never said that it was a problem, nor did I say I was expecting something different, if you reread my post all I did was mention that to me, mounting a 50mm lens on a camera and cranking it wide open to F1.0 but getting a F1.4 focal plane is what I consider "quirky" aka not typical.....in other words using a speedbooster in general is not something that I am used to but it is what is needed on the C70 if you want a FF FOV....that literally was all that statement meant.

 

[M_Williams]

8 hours ago, herein2020 said:

No it really seems like you just didn't understand my post at all....I never said that it was a problem, nor did I say I was expecting something different, if you reread my post all I did was mention that to me, mounting a 50mm lens on a camera and cranking it wide open to F1.0 but getting a F1.4 focal plane is what I consider "quirky" aka not typical.....in other words using a speedbooster in general is not something that I am used to but it is what is needed on the C70 if you want a FF FOV....that literally was all that statement meant.

 

Ok fair enough. You said you found it "strange" which to me means "out of the ordinary" but I understand what you mean. It is quirky especially if you use a light meter like I still do.

[herein2020]

I tested out the new C70 firmware today and shot in Cinema RAW LT at 30FPS. On a whim, I directed the RAW footage to record to my V30 card and it actually recorded fine.

I just shot a few seconds of footage to test out playback on my editing system and to see if my SD cards could handle it. I definitely think my 1TB SandDisk Extreme Pro cards are rock solid, it was almost comical seeing 411 min of recording time available when recording to the 1TB card in RAW LT. 

When I tested RAW LT at 60FPS though it buffer overflowed immediately. So my storage concerns were definitely unwarranted; this camera, despite its faults is really turning into a rock solid workhorse for me. Below is a quick table that I put together to show the codecs that I care about and their recording times.

 

• SanDisk Extreme Pro 1TB V30
  • Canon RAW LT -> 29.97FPS -> Recording Time 422min
  • Canon XF-AVC Long GOP -> 29.97FPS -> Recording Time 852min
  • Canon XF-AVC Long GOP -> 59.97FPS -> Recording Time 524min
• ProGrade 256GB V90
  • Canon RAW ST -> 29.97FPS -> 68min
  • Canon RAW LT -> 29.97FPS -> Recording Time 105min
  • Canon RAW LT -> 59.97FPS -> Recording Time 52min
  • Canon XF-AVC Long GOP -> 29.97FPS -> Recording Time 213min
  • Canon XF-AVC Long GOP -> 59.97FPS -> Recording Time 131min

Also, the Canon RAW LT at 29.97FPS seems to have a bitrate of 322Mb/s which is a data rate of 40MB/s, which is only half the 80MB/s that my SanDisk V30 card registered during its tests so I should be pretty safe with it. I will definitely keep the ProGrade cards in my bag, but it looks like for 90% of my work I am fine with keeping two 1TB SanDisk Pro cards in the camera with it set to dual slot recording.

If I shoot a project that requires raw footage I will probably just use two ProGrade cards, not worth the headache of trying to remember which footage gets recorded to where depending on framerate.

[ntblowz]

C70's raw file size is much more manageable than c200/c300's 1gbps monster and that's only 10bit in 4k50/60p vs 12bit on c70.

 

The 2k raw is even usable in wedding too lol (for ceremony, not sure about reception cause of lowlight on s16 crop).

 

This raw update really makes c70 attractive! No need bulky external monitor + battery if you want compact raw cinema camera.

 

[herein2020]

7 hours ago, ntblowz said:

C70's raw file size is much more manageable than c200/c300's 1gbps monster and that's only 10bit in 4k50/60p vs 12bit on c70.

 

The 2k raw is even usable in wedding too lol (for ceremony, not sure about reception cause of lowlight on s16 crop).

 

This raw update really makes c70 attractive! No need bulky external monitor + battery if you want compact raw cinema camera.

 

 

That is a good point, I haven't looked at the cropped modes at all except when testing crop sensor lenses. BTW in cropped mode APS-C lenses work fine so that opens up many more options for cheap lenses when combined with the straight through adapter or speedbooster.

I don't think cropped mode would affect its lowlight capabilities; sure, it is using less of the sensor, but it is also producing a lower resolution image so the pixel density is the same. I think the bigger question would be if the cropped FOV would be wide enough to still be useable. Cropped mode with RAW LT  combined with my APS-C Sigma 18-35 F1.8 and the speedbooster dropping that to F1.4 could be an interesting combination for shooting RAW in lowlight for something like a wedding reception.

[TomTheDP]

13 hours ago, herein2020 said:

 

That is a good point, I haven't looked at the cropped modes at all except when testing crop sensor lenses. BTW in cropped mode APS-C lenses work fine so that opens up many more options for cheap lenses when combined with the straight through adapter or speedbooster.

I don't think cropped mode would affect its lowlight capabilities; sure, it is using less of the sensor, but it is also producing a lower resolution image so the pixel density is the same. I think the bigger question would be if the cropped FOV would be wide enough to still be useable. Cropped mode with RAW LT  combined with my APS-C Sigma 18-35 F1.8 and the speedbooster dropping that to F1.4 could be an interesting combination for shooting RAW in lowlight for something like a wedding reception.

It definitely will IMO. Full frame vs S35 on say the Panasonic S1 is dramatically different if you are struggling for light. You are essentially blowing up the noise. If you expose correctly its no issue of course.

[kye]

On 3/26/2022 at 12:38 AM, herein2020 said:

I tested out the new C70 firmware today and shot in Cinema RAW LT at 30FPS. On a whim, I directed the RAW footage to record to my V30 card and it actually recorded fine.

I just shot a few seconds of footage to test out playback on my editing system and to see if my SD cards could handle it. I definitely think my 1TB SandDisk Extreme Pro cards are rock solid, it was almost comical seeing 411 min of recording time available when recording to the 1TB card in RAW LT. 

When I tested RAW LT at 60FPS though it buffer overflowed immediately. So my storage concerns were definitely unwarranted; this camera, despite its faults is really turning into a rock solid workhorse for me. Below is a quick table that I put together to show the codecs that I care about and their recording times.

 

• SanDisk Extreme Pro 1TB V30
  • Canon RAW LT -> 29.97FPS -> Recording Time 422min
  • Canon XF-AVC Long GOP -> 29.97FPS -> Recording Time 852min
  • Canon XF-AVC Long GOP -> 59.97FPS -> Recording Time 524min
• ProGrade 256GB V90
  • Canon RAW ST -> 29.97FPS -> 68min
  • Canon RAW LT -> 29.97FPS -> Recording Time 105min
  • Canon RAW LT -> 59.97FPS -> Recording Time 52min
  • Canon XF-AVC Long GOP -> 29.97FPS -> Recording Time 213min
  • Canon XF-AVC Long GOP -> 59.97FPS -> Recording Time 131min

Also, the Canon RAW LT at 29.97FPS seems to have a bitrate of 322Mb/s which is a data rate of 40MB/s, which is only half the 80MB/s that my SanDisk V30 card registered during its tests so I should be pretty safe with it. I will definitely keep the ProGrade cards in my bag, but it looks like for 90% of my work I am fine with keeping two 1TB SanDisk Pro cards in the camera with it set to dual slot recording.

If I shoot a project that requires raw footage I will probably just use two ProGrade cards, not worth the headache of trying to remember which footage gets recorded to where depending on framerate.

Most compressed codecs will have a lower bitrate when there isn't huge amounts of movement.

I'd suggest a stress test to max out the codec.  One I have done before is to get half-a-dozen stills and put them on a 60p timeline at a single frame each, the point the camera at your monitor so that the whole frame is moving, and turn all the lights out, then set the computer to loop on that timeline, and hit record on a 24/25/30p mode.

This ensures that no two frames will be remotely similar and it should stress the codec and push the camera to record the maximum bitrate for that codec.  I've seen increases in bitrate by even 50% when doing this, compared to other test footage.  I've tried pointing the camera at trees moving in the wind and this laptop test stresses the codec more, and is available even if it isn't windy and if you don't have any trees nearby 🙂 

[BenEricson]

I think the main issue with these Canon cameras is still the post work flow. There is still no Cinema Gamut to REC.709 LUT. The conversion LUT from Log2 to Wide DR doesn't even match the baked in Wide DR profile. (I've tried. They don't match.)

The Komodo for example, looks MUCH better right out of the gate with the standard RED provided LUT applied. There are 4 different LUTs with 4 different roll off curves for you to choose. You can preview whatever contrast level you want in camera, then apply the correct LUT in your NLE. It looks exactly like it does in camera.

The LT raw footage played back pretty well for me. The other codecs work smoother though. I'm on an older Mac Pro. I don't really notice a difference in the footage though.

[herein2020]

On 3/26/2022 at 11:46 AM, TomTheDP said:

It definitely will IMO. Full frame vs S35 on say the Panasonic S1 is dramatically different if you are struggling for light. You are essentially blowing up the noise. If you expose correctly its no issue of course.

I guess I just don't understand why that would be. If you are using the same number of photosites on the sensor for a given resolution how could there be more noise since you are not increasing the gain when cropping in? It would make more sense to me if say the native ISO increased when switching to the S16 mode, but without an increase in gain how is there an increase in noise? Maybe the S1 line skips when in cropped mode or some maybe some kind of upscaling is occurring which decreases the image quality or maybe its circuits are just more noisy when handling the crop mode.

But with the C70 in cropped mode and RAW there is no line skipping or upscaling, it is just using a smaller region of the sensor and the native ISO does not change. It would be very interesting to see someone perform a noise test for the C70 in the cropped vs S35 mode. 

 

15 hours ago, kye said:

Most compressed codecs will have a lower bitrate when there isn't huge amounts of movement.

I'd suggest a stress test to max out the codec.  One I have done before is to get half-a-dozen stills and put them on a 60p timeline at a single frame each, the point the camera at your monitor so that the whole frame is moving, and turn all the lights out, then set the computer to loop on that timeline, and hit record on a 24/25/30p mode.

This ensures that no two frames will be remotely similar and it should stress the codec and push the camera to record the maximum bitrate for that codec.  I've seen increases in bitrate by even 50% when doing this, compared to other test footage.  I've tried pointing the camera at trees moving in the wind and this laptop test stresses the codec more, and is available even if it isn't windy and if you don't have any trees nearby 🙂 

That is true, I do need to do more testing, my favorite test is a waterfall where nearly every pixel changes from one frame to the next.  I don't have the patience to do the whole computer monitor stills thing 🙂 I am not a very good tester in general, I'd much rather be out shooting than testing. I work through the basics then get out there and shoot.

 

14 hours ago, BenEricson said:

I think the main issue with these Canon cameras is still the post work flow. There is still no Cinema Gamut to REC.709 LUT. The conversion LUT from Log2 to Wide DR doesn't even match the baked in Wide DR profile. (I've tried. They don't match.)

The Komodo for example, looks MUCH better right out of the gate with the standard RED provided LUT applied. There are 4 different LUTs with 4 different roll off curves for you to choose. You can preview whatever contrast level you want in camera, then apply the correct LUT in your NLE. It looks exactly like it does in camera.

The LT raw footage played back pretty well for me. The other codecs work smoother though. I'm on an older Mac Pro. I don't really notice a difference in the footage though.

For the work that I do, what comes out of the C70 is good enough for me. I agree though, the back screen always looks perfect and I have yet to find a LUT that can reproduce that look in post. There is a great YT video that I posted earlier in the thread where he went through many of the available LUTs and showed the results. I ended up going with the Buttery LUT for my Rec709 workflow, it looks the most natural to me and gets me the closest to where I want to be.

[kye]

7 hours ago, herein2020 said:

That is true, I do need to do more testing, my favorite test is a waterfall where nearly every pixel changes from one frame to the next.  I don't have the patience to do the whole computer monitor stills thing 🙂 I am not a very good tester in general, I'd much rather be out shooting than testing. I work through the basics then get out there and shoot.

Hahaha...  So, packing up camera gear, driving to a waterfall, carrying gear to the waterfall, <shooting>, and then packing the gear up, trekking back to the car, driving back to your house, and unpacking your gear, somehow all take a negative amount of time?

The <shooting> part in the middle is the same, and putting a few stills on a timeline is literally 2 minutes of work.

[herein2020]

51 minutes ago, kye said:

Hahaha...  So, packing up camera gear, driving to a waterfall, carrying gear to the waterfall, <shooting>, and then packing the gear up, trekking back to the car, driving back to your house, and unpacking your gear, somehow all take a negative amount of time?

The <shooting> part in the middle is the same, and putting a few stills on a timeline is literally 2 minutes of work.

Oh I would never do that just to test the camera, I just like shooting waterfalls since there are none where I live; nearest waterfall is probably a 10hr drive. I should have better worded my response. Waterfalls are where I have seen compressed codecs fall apart the fastest so if I had a chance to shoot a waterfall I think that's where the data rate for the codec would be the highest.....so to me that would be the best real world test of the RAW LT data rates using a subject that I actually shoot from time to time vs using the monitor setup.

As far as actually testing the codec specifically, my way of testing it is just getting out there and shooting. For the first few paying gigs I will just keep a V90 in one of the slots just like I did for XF-AVC.

To me, sure in a controlled test environment it may be possible to get it to buffer overflow using the monitor test, but if it never does so in the real world shooting the projects that I shoot which is not fast action or that much difference from frame to frame then for my purposes my setup meets my needs.

[kye]

2 hours ago, herein2020 said:

Oh I would never do that just to test the camera, I just like shooting waterfalls since there are none where I live; nearest waterfall is probably a 10hr drive. I should have better worded my response. Waterfalls are where I have seen compressed codecs fall apart the fastest so if I had a chance to shoot a waterfall I think that's where the data rate for the codec would be the highest.....so to me that would be the best real world test of the RAW LT data rates using a subject that I actually shoot from time to time vs using the monitor setup.

As far as actually testing the codec specifically, my way of testing it is just getting out there and shooting. For the first few paying gigs I will just keep a V90 in one of the slots just like I did for XF-AVC.

To me, sure in a controlled test environment it may be possible to get it to buffer overflow using the monitor test, but if it never does so in the real world shooting the projects that I shoot which is not fast action or that much difference from frame to frame then for my purposes my setup meets my needs.

Fair enough.

Use a long lens and fill the frame with extreme motion, ideally the water in one frame should be completely gone in the next frame, otherwise the compression can use the top part of one frame to describe the bottom part of the next frame.  Also, remove any NDs and set it to the fastest SS you can muster to get the most detail in the image and make the frames more different to each other (a blur in one frame looks a lot like a blur in the next frame).

The idea of stress testing these things is so that you never hit a situation where the equipment will fail on you.  The last thing you want is to have some situation you didn't anticipate happen and then learn the hard way that the card doesn't work.  Surprises happen - you might think you're just shooting people who aren't moving much, but then a woman shows up in a dress that's got fine texture with sparkles and every frame is completely different from the previous one with tiny little movements and then the card stops working and you're left in the edit suite missing footage and wondering how to recover.

[kye]

10 hours ago, herein2020 said:

I guess I just don't understand why that would be. If you are using the same number of photosites on the sensor for a given resolution how could there be more noise since you are not increasing the gain when cropping in? It would make more sense to me if say the native ISO increased when switching to the S16 mode, but without an increase in gain how is there an increase in noise? Maybe the S1 line skips when in cropped mode or some maybe some kind of upscaling is occurring which decreases the image quality or maybe its circuits are just more noisy when handling the crop mode.

But with the C70 in cropped mode and RAW there is no line skipping or upscaling, it is just using a smaller region of the sensor and the native ISO does not change. It would be very interesting to see someone perform a noise test for the C70 in the cropped vs S35 mode. 

Cameras that downsample will benefit from that as the downsampling reduces the noise, compared to punching-in to a 1:1 readout.  If you're shooting RAW then there wouldn't be so much of an advantage, except that the grain of the noise will be finer at higher resolutions.  

A pixel from S16 2048x1080 will be 2.93x the size of a 6K 6008x3168 pixel when you stretch both of the clips to the full timeline resolution.