Pavel Sedlak
March 10th, 2014, 06:07 AM
The NoiseReduction in camera test (NR0, NR2, NR4),
AVCHD vs Ninja2 in poor light scene.
C100 and Ninja2 in camera noise reduction test (cine1, wdr, canon-log gamma, 25p and 25i) on Vimeo
Samyang 24mm/f1.5, a very poor light, ISO850, ISO1600, ISO4000, internal AVCHD+external Ninja2 recording.
Avid Symphony at postproduction, I was working with these versions of clips (in this order):
1. AVCHD in XDCAM long GOP HQ 35Mbps 420 codec
2. AVCHD in Avid DNxHD185 (422, 8bit. I-Frame only) codec
3. Ninja2 version in Avid DNxHD185 (true 422, 8bit, I-Frame only) codec
4. repeat "1." for better visual contrast with "3."
All these codecs are present with NR0, NR2 and NR4 versions, with three levels of ISO. There were NO COLOR CORRECTIONs at post, I did only a minor black setup (with master pedestal) for the same start point in black and I increased (a lot) the gain level for canon-log clips with low ISO, nothing else (in higher ISO there was no need for such increasing level of gain in post, the canon-log result is much better in this case).
The result was exported in DNxHD185 (422, 8bit) and for vimeo this master file was transcoded in Sorenson Squeeze to H264, with 15Mbps in slow (high quality) rendering. I sugest to download the orig H264 file (DNxHD 185Mbps version is the best, but it is too big file for vimeo).
C100 has a higher noise level in black parts of pictures, the higher ISO level helps with this so the best result has cine1 gamma, NR2 and ISO4000 level version of clips (in this case of scene with a poor light). Also different gamma have a different noise levels (and 25p version is much sharper and better than 25i version).
In the picture is also a small PIP with the 400pct magnification which helps to see the differences (also in chroma subsampling 420 vs. 422 on the test chart). ABB was made for all different ISO parts. Chroma subsampling differences are also visible for example on the brown paw of the rabbit (22:43).
Thanks to Pavel LOPOLO Jirak for lending the Ninja2 recorder.
(the best part is at 19:00, the worse is at 18:00).
AVCHD vs Ninja2 in poor light scene.
C100 and Ninja2 in camera noise reduction test (cine1, wdr, canon-log gamma, 25p and 25i) on Vimeo
Samyang 24mm/f1.5, a very poor light, ISO850, ISO1600, ISO4000, internal AVCHD+external Ninja2 recording.
Avid Symphony at postproduction, I was working with these versions of clips (in this order):
1. AVCHD in XDCAM long GOP HQ 35Mbps 420 codec
2. AVCHD in Avid DNxHD185 (422, 8bit. I-Frame only) codec
3. Ninja2 version in Avid DNxHD185 (true 422, 8bit, I-Frame only) codec
4. repeat "1." for better visual contrast with "3."
All these codecs are present with NR0, NR2 and NR4 versions, with three levels of ISO. There were NO COLOR CORRECTIONs at post, I did only a minor black setup (with master pedestal) for the same start point in black and I increased (a lot) the gain level for canon-log clips with low ISO, nothing else (in higher ISO there was no need for such increasing level of gain in post, the canon-log result is much better in this case).
The result was exported in DNxHD185 (422, 8bit) and for vimeo this master file was transcoded in Sorenson Squeeze to H264, with 15Mbps in slow (high quality) rendering. I sugest to download the orig H264 file (DNxHD 185Mbps version is the best, but it is too big file for vimeo).
C100 has a higher noise level in black parts of pictures, the higher ISO level helps with this so the best result has cine1 gamma, NR2 and ISO4000 level version of clips (in this case of scene with a poor light). Also different gamma have a different noise levels (and 25p version is much sharper and better than 25i version).
In the picture is also a small PIP with the 400pct magnification which helps to see the differences (also in chroma subsampling 420 vs. 422 on the test chart). ABB was made for all different ISO parts. Chroma subsampling differences are also visible for example on the brown paw of the rabbit (22:43).
Thanks to Pavel LOPOLO Jirak for lending the Ninja2 recorder.
(the best part is at 19:00, the worse is at 18:00).