C100: C500 and C100 cine1 vs canon-log test

P

Pave1

Well-known member
There was a Canon day in Prague (Czech republic) and I could test the C500 and C100 (with Canon cine-primes, I chose the 24mm and 50mm lens).

http://vimeo.com/87787323 (only quick color corrections)

The C500 was with the internal long GOP 422 8bit and external "I-Frame only" 422 8bit record (we hadn't 3G-SDI recorder, only HD-SDI, so C500 was in "C300 mode"), the C100 was with the internal AVCHD 420 8bit record, all ISO 850, no NR in camera (or at post).

Canon cine-primes 24mm and 50mm, postproduction was with Avid Symphony (DNxHD185 422 codec, a quick color correction of canon-log only on my notebook screen), an export was with the Sorenson Squeeze encoder (x264, 15Mbps). There is also 25p and 25p* (*from 25i with deinterlace) comparison.

The C100 was new with only one ABB, so the noise level could be better (we tried this in the second day, comparison is at the end). My personal C100 is clear in this ISO setting.

Thanks to Canon.cz for C500.
 
thank you so much for this. Been wanting to see a C100 side by side with a C500. They look nigh on which is amazing considering the price difference. And when you factor in an external recorder - the C100 becomes a hell of camera for the money.
 
Thanks. We have C500 only with HD-SDI externall record (C500 was in "C300 mode" in HD with only 8bit 422 output), so we can't to test the 10bit 422 and 12bit 444 HD or 2K/4K raw record. It is clear that C500 has much more type of output than C300 or C100, we test only one of them (internal and external).

C100 is nice camera, I made test of workflow in Avid Symhony/Composer 6.5, differences between cine1 and canon-log, between 25p and 25i deinterlaced (with slow Avid fluid morphing), and I finish also the codec test now - working with AVCHD from C100 in XDCAM EX HQ 35Mbps long GOP (renders in 185Mbps) vs uncompressed DNxHD 1:1 (renders in 185Mbps). There are different noise levels, differences in details or in smoothing of gradients.
 
Here is the version from uncompressed (Avid Symphony DNxHD 1:1) AVCHD/XF/ProResHQ clips.
https://vimeo.com/87923721

AVCHD from the C100 in the first version was in XDCAM EX HQ 35Mbps 420, now is in DNxHD 1:1 422.
C500 XF codec and extrenal ProResHQ are in Avid Symphony decompressed to DNxHD 1:1 422.

All clips has less or more color correction and full processing (titles, all are processed*),
the export was in DNxHD185 8bit (185Mbps, 422, 8bit) and squeezed in Sorenson encoder
to the H264 (x264, 15Mbps).

I can see some differences in both version, in cine1 (AVCHD) is better noise because there is less compression artefacts,
also smooth gradient on the wall is interesting in canon-log C100 version.
 
Last edited:
This kinda misses the point of C500. It's strength is in 12-bit 444 or RAW. (I know you said that it was internal only). Even though, I notice quite a difference in the amount of noise. That bit-rate makes a difference.
 
as i wrote, we hadn't 3G-SDI recorder, only HD-SDI, so C500 was in "C300 mode". I hope that next week we can try full C500 power.

For me is more interesting how well is performed smooth gradient on the wall than noise level at dark places because cameras were new (with only one or two auto black balance, so noise level can be better). I had to do the basic menu setup on both cameras .-) .
 
ABB has a lot with noise, because during this calibration of the chip ABB also read the noise for better denoiser function (there is denoiser at early stage on analogue signal on the CMOS chip, you can look at this info in Canon white paper) . There are three types of noise, but I can't tell you more, because I'm not an engineer from Canon compeny. But I really think that several ABB in case of the new cam is better, because the noise reading is changing during the time, ISO levels, etc.

I did ABB on my own C100 repeatedly (with changing ISO, 25p vs 25i), believe me or not .-) .

My C100 is with the cine1 gamma clean to ISO 5000, so I think that more ABB can help with this.
 
Last edited:
Thanks Pave1.

I expect your test may well reveal subtle differences in the cutting room and may translate to quantifiable differences on a cinema screen.

For me, it further confirms that for those of us publishing exclusively to the web, with a transcode from master to H264 and subsequent transcoding by Vimeo or similar, the differences to the casual viewer between the different acquisition codecs is indistinguishable.

Interesting that key influencers in the BBC/EBU are coming round to the idea that the latest implementation of MPEG4 codecs (eg C100) really are visually indistinguishable from the relatively ancient and inefficient MPEG2 50 Meg codecs (eg C300/500 which exist primarily to tick the min 50 Meg box.) And, crucially, they hold up just as well through the post production/publishing chain too.

Ben.
 
bengiles said:
...I expect your test may well reveal subtle differences in the cutting room and may translate to quantifiable differences on a cinema screen.
Click to expand...

Yes, the primary purpose was to see the subtle differences between 8bit vs 10bit type of recording, but we hadn't the right recorder, so we did the 8bit signal test only.

Signal output in 185Mbps 422 8bit (DNxHD185, mov) showed these subtle differences (eg on the smooth gradient on the wall) in the very clean form, but re-export to the H264 showed how well the signal work after another processing. This part is interesting for H264 otput to the web and also for broadcast purposes, because at the end is output from the broadcast master transcoded to the distribution type of compression (with highend type of coder). Another level of the test is to see this result on different type of TV (plasma, LCD TV) with its own different types of the internal processing.

There are visible subtle differences between 25p and 25i deinterlaced modes (in the sharpness and also on the smooth gradient at the wall), in 25p mode there are differences in noise level in black part (what is important for grading) and banding (and colors) in the smooth gradient on the wall. 422 from the C500 (in C300 8bit mode) is more robust in canon-log gamma, where AVCHD looks worse. C300 in cine1 is not far from AVCHD, in 8bit has cine1 gamma on C100 a very natural looking image with nice color.
Interesting are also subtle differences in the uncompressed vs DNxHD185 processing in Avid Symphony, where in the master export file in DNxHD185 is uncompressed file better but in the other processing (from 8bit to H264) is better DNxHD185 processing, because a very subtle compression artefacts help with smoothing of the gradient (only 8bit) on the wall in the H264 result. 25i (deinterlaced) version is in this H264 result worse.

You must also see that this type of scene has no moving, differences between AVCHD and XF codec (422 long GOP, 50Mbps) can be bigger in another type of scene, also 25i vs 25p can get a different result in the quick move. This test showed differences in basic DR (cine1 vs canon-log), in the 8bit signal limitation (in the smooth gradient), in noise level at black part of images (input quality for color grading). I also want to see the differences in quick type of workflow or with higher ISO in cine1, this was the reason why I used the ISO850 with cine1 gamma curve instead of ISO320 and why I tried to used both the DNxHD185 codec and uncompressed signal in Avid Symphony.

All these differences are a very subtle, but if you know them you can get the better result in complex scenes (with or without moving, in 25p or 25i, with cine1 or canon-log gamma, with a quick or slow uncompressed workflow, etc.).
 
Last edited:
You must log in or register to reply here.
Back
Top