S-LOG vs SXS, which is cleaner in low light?
- Thread starter Lkorver
- Start date
ProRes is better than DNxHD- bit more efficient and less tends to smooth and block. It also better preserves easy frames of the footage.
Not sure what people mean by noisy?
Duke M.
Veteran
Reports from Gemini users were that ProRes has many more artifacts than uncompressed. I think that almost goes without saying.
I will defer to your expertise but my expectation based on minor comparisons is that ProRes HQ and the best quality DNxHD would yield very similar results with only minor trade offs, though DNxHD would be a little bigger file.
Last edited:
Yes- they are similar, but ProRes has some edge. Also- it's obvious that they will have some difference than uncompressed, but this should never be seen at 1:1 frame. You need to zoom at least 3x to see difference. If it's not the case than it may be due to Atomos implementation- but they claim Apple had to approve it.
Check this (page 13):
http://images.apple.com/support/finalcutpro/docs/Apple-ProRes-White-Paper-July-2009.pdf
and this (page 8):
http://www.grassvalley.com/docs/WhitePapers/professional/GV-4097M_HQX_Whitepaper.pdf
Both tests (independent and different) confirm each other- I also have STEM footage (which has different section to check all weaknesses of codecs) and can confirm ProRes numbers. I also deal a lot with DNxHD footage and seen it going into blocking on difficult, noisy, specially interlaced sources.
Someone needs to take uncompressed recording and convert to ProRes in eg. FPC and compare to Atomos conversion (eg play it over HD-SDI/HDMI and reocord to Atomos).
One important thing- this should be raw camera footage, which has not went through any compression before!
Had finally a chance to looks at F3 footage and play with camera at BVE.
I've also seen other cameras- Panasonic, Alexa, Canon....
The cleanest of F3 is simply amazing. Color accuracy is also spot on in 709, without any tweaking- looking at real world and Sony broadcast monitor was a joy- all matched simply perfectly. I was also playing with their new zoom lens- very impresive. There is something (of course it's probably sensor), which make F3 footage stand out of others. All details are so well defined, even zooming into dark areas of the roof hall still provided clean and very well defined image. All edges were so smooth (camera was connected through dual SDI to Sony nice broadcast monitor- for note) and of course almost no noise. In the same time camera was on default settings, there was no additional/setup lighting (like it was on Panasonic, Canon stand). I was simply amazed by F3 footage- then went to Panaosinc, Canon stand and....no, none of their cameras where no near F3, even Canon C300 wasn't that impressive- I was actually dissapointed by the look if it (even when it was setup with brand new Canon massive zoom lens). I think there was some problem with the actual setup (visitors mess around with setting etc , so easy to break). Panasonic cameras were ok- some better, some worse, but again- none of them stand out- they look very video like. F3, without any lighting/setup just stands out a lot. I think F3 sensors is great- especially when Sony decided to make camera only HD, even if sensor is much bigger. This is for me the reason for this nice look and simply amazing cleanest of the image. Even if Sony new zoom lens is not very fast still looks very good and it's sharp to very end of range.
Sony new 8K camera is yet another story. Only one other camera which did look very good in every situation was Alexa- this is also great piece of hardware. Same story- about 4K sensor to deliver just HD image- this is the reason behind F3 and Alexa "special" look I think.
I've also seen other cameras- Panasonic, Alexa, Canon....
The cleanest of F3 is simply amazing. Color accuracy is also spot on in 709, without any tweaking- looking at real world and Sony broadcast monitor was a joy- all matched simply perfectly. I was also playing with their new zoom lens- very impresive. There is something (of course it's probably sensor), which make F3 footage stand out of others. All details are so well defined, even zooming into dark areas of the roof hall still provided clean and very well defined image. All edges were so smooth (camera was connected through dual SDI to Sony nice broadcast monitor- for note) and of course almost no noise. In the same time camera was on default settings, there was no additional/setup lighting (like it was on Panasonic, Canon stand). I was simply amazed by F3 footage- then went to Panaosinc, Canon stand and....no, none of their cameras where no near F3, even Canon C300 wasn't that impressive- I was actually dissapointed by the look if it (even when it was setup with brand new Canon massive zoom lens). I think there was some problem with the actual setup (visitors mess around with setting etc , so easy to break). Panasonic cameras were ok- some better, some worse, but again- none of them stand out- they look very video like. F3, without any lighting/setup just stands out a lot. I think F3 sensors is great- especially when Sony decided to make camera only HD, even if sensor is much bigger. This is for me the reason for this nice look and simply amazing cleanest of the image. Even if Sony new zoom lens is not very fast still looks very good and it's sharp to very end of range.
Sony new 8K camera is yet another story. Only one other camera which did look very good in every situation was Alexa- this is also great piece of hardware. Same story- about 4K sensor to deliver just HD image- this is the reason behind F3 and Alexa "special" look I think.
tom.wong
Well-known member
the thing about s log is that it uses gain to get that extra highlight retention. if you look at the typical curve without s log, you retain more on your step chart on the lower end than the top end. when you increase the iso it starts evening out. for all intens and purposes, s log isn't just that hyper flat image to give you more image information, the gain gives you that highlight retention too. the noise at on s log at 1600 iso should be identical to 1600 iso without s log. noise is noise, the sensor always produces the same amount. a gamma curve won't change that. it's all about the design of the sensor and the processing. no gamma curve can change how much noise the sensor produces.
going to sxs however, puts up a lot of compression artifacts like macro blocking. it can be very clean too but having so much information crammed in a 8 bit container causes a lot of issues when you really want to dig into the image. 8 bit vs 10 bit discussed to death.
pro res isn't a lossless codec, it's a high quality compressed codec. and can be fairly lossy. direct comparisons against even hdcam sr and cineform shows pro res as the loser. same vs uncompressed of course. it's just that pro res is industry accepted, easy to edit with, and manageable file sizes.
going to sxs however, puts up a lot of compression artifacts like macro blocking. it can be very clean too but having so much information crammed in a 8 bit container causes a lot of issues when you really want to dig into the image. 8 bit vs 10 bit discussed to death.
pro res isn't a lossless codec, it's a high quality compressed codec. and can be fairly lossy. direct comparisons against even hdcam sr and cineform shows pro res as the loser. same vs uncompressed of course. it's just that pro res is industry accepted, easy to edit with, and manageable file sizes.