FS5 Frame Rates

[John Gilmore]

I've just got my FS5 and I was wondering about what frame rate to use.

I intend to shoot XAVC HD, as QFHD gives only 4:2:0.

My question is what frame rate will give the best quality?

I have seen recomendations which state that 1080/25p 50Mbps will give the best quality.

So I have two questions.

Firstly, should I use the NTSC or PAL frame rates? As I'm based in the UK, I assume that the 50Hz (PAL) frame rates would be the natural choice, although you do seem to get more options with the NTSC 60Hz frame rates.

Secondly,assuming a bit rate of 50Mbps, will I sacrifice image quality by using the higher frame rates (60p/50p) or will I obtain better quality by using the lower frame rates (24p/25p)?

[Jody Arnott]

Using 50p is like stuffing twice the amount of frames into the same sized container (bitrate), so some extra compression is going to have to occur which might sacrifice image quality. I've never noticed a difference between 25/50p personally, but I generally stick to 25p unless I need slow motion. 25p is also easier to work with in my NLE.

As for PAL/NTSC, I don't think it matters too much if you are shooting for the web. If you're shooting for broadcast purposes, it's a whole different story. I shoot for the web and stick to PAL purely because I live in a PAL part of the world... plus it's a habit from my days working in broadcast.

[Cliff Totten]

If the FS5's UHD 4:2:0 chroma sampling vs. it's 1080 4:2:2 sampling is your biggest concern, it can be argued that UHD 4:2:0 contains more (or at least as much) chroma resolution as 1080 4:2:0. (especially on a single Bayer sensor)

If you shoot UHD 4:2:0 for the purpose of 1080 4:2:2 delivery, you can certainly achieve that when you render that out to 1080. And I think we can all agree that a UHD capture renders down to a fantastic 1080 image. In my opinion, it's MUCH better than a native 1080 capture.

Now, a 1080 10bit color depth vs an 8 bit UHD color depth is a stronger argument in a heavy color grading work flow.

CT

[John Wiley]

One thing often overlooked with inter-frame codecs is that when the frame rate increases, the amount of difference between frames decreases. Hence the amount of data required to record the per-frame changes is lower. This means the difference in quality is a lot less than the "double the bit-rate per frame" argument in favour of 25p.

My advice would be to use whatever the situation calls for. If you're delivering in 25p and you know you're not going to be needing slow motion (eg: an interview) then stick to 25p. If you might want to use slow-mo or don't know the delivery format yet, just use 50p without stressing about quality loss.

[Nigel Davey]

You may find this older thread on the same topic useful: http://www.dvinfo.net/forum/sony-xdc...0p-50-60p.html

[John Gilmore]

I’ll be using the camera on my forthcoming trip to the Northwest Passage, so it will mainly be travel documentary style consisting of landscape plus on-board footage of ship life etc.

The footage is more for personal use and I’ll probably end up cutting a Blu-ray and just watching it at home occasionally.

I probably won’t do much in post-production, although I may use it as a learning exercise.

My main reason for considering 50p is that I thought the footage would be smoother.

Similarly, I imagine the 10-bit colour depth of HD would produce better results than QFHD’s 8-bit depth.

[John Gilmore]

Originally Posted by Nigel Davey

You may find this older thread on the same topic useful: http://www.dvinfo.net/forum/sony-xdc...0p-50-60p.html

An interesting thread, thanks for that Nigel.

So are you still using 50p or did it swing you over to 25p?

[Nigel Davey]

Most of the time I shot in 50p in HD. But that's because quite a lot of what I do gets cut (partly) to music and the slomo option is always nice to have up your sleeve, particularly for limited time ramping.

But if I'm doing interviews I'll often shoot in 4k 25p to allow re-framing in Post.

But what I did discover ages ago is that inside Premiere Pro you should avoid having 50p sequences with 50p footage on them (ie don't generate a sequence from a 50p clip). Although this seems the logical thing to do it will result in all your render times being approx twice as long at a manually set (or adjusted) 25p sequence with 50p clips on them (where Premiere has thrown away half the frames to get down to 25p). Obviously this approach only applies if you are exporting at 25p (or 30p across the pond).

[Cliff Totten]

Originally Posted by John Gilmore

I’ll be using the camera on my forthcoming trip to the Northwest Passage, so it will mainly be travel documentary style consisting of landscape plus on-board footage of ship life etc.

The footage is more for personal use and I’ll probably end up cutting a Blu-ray and just watching it at home occasionally.

I probably won’t do much in post-production, although I may use it as a learning exercise.

My main reason for considering 50p is that I thought the footage would be smoother.

Similarly, I imagine the 10-bit colour depth of HD would produce better results than QFHD’s 8-bit depth.

Northwest Passage? Sounds like some epic scenery. If it were me, I wouldn't hesitate in shooting in 4k.

As far as the 8bit vs 10bit argument goes, if you are shooting SLOG, than certainly 10 bits will protect you against color banding. But, if you are shooting any rec709-ish profile where 95% of your image is already there, those 8bits will certainly survive light grading with no problem. Generally speaking, I will take an 8bit 4k image over a softer and lower resolution 10bit 1080 image anyday. (unless heavy grading is needed)

Aslo,...how do I say this? A 4:2:0, 8bit h.264 file at 50Mbp/s is practically bulletproof compression. But,....change that to 4:2:2 (double the chroma resolution) and increase it 10 bit (increasing the color depth by orders of magnitude)....and suddenly your compression ratio shoots up. Sony could have created a very nice 100Mbp/s, 10bit 1080 mode because we know that SDXC cards can easily do that bitrate. But Sony suspiciously left that option out.

Sony seems to want a very BIG CODEC "gap" between all the codecs shot on cheap SDXC cards which it doesnt make any money on and XQD cards which are VERY expensive and licensed and owned by Sony. It seems that higher bitrate long GOP CODECS make intra-frame CODECS on XQD cards LESS necessary. This is not a good thing for XQD cards. So,.....they leave the SDXC CODECS extra, extra, extra, low and try to blame it on the speed of SDXC cards. (meanwhile, JVC, Panisonic, BlackMagic and others EASILY record way over 100Mbp/s on their SDXC cards) Go figure....;-)

Anyhoo, if you haven't played with 4k yet. Go out to your backyard and shoot two scenes, one in 1080 10bit and again in UHD 8bit. Go back inside see the difference. It's my guess that you will like the UHD 8bit much better. Remember, Blu-ray is only 8bit and most consumer TV/monitors today are still only 8 bit.

CT

[John Gilmore]

Thanks Cliff, that's very useful information.

I now intend to shoot in QFHD, especially since the Blu-ray standard (as you observed) allows for a maximum colour depth of 8 bits.

Anyway, it seems I have two options for 2160/25p: 60Mbps and 100Mbps.

Can anyone shed any light on the downsides and upsides of each of these options?

[Gary Huff]

Originally Posted by John Gilmore

Anyway, it seems I have two options for 2160/25p: 60Mbps and 100Mbps.

60Mbps will be smaller, but the quality may or may not be noticeably worse. 100Mbps will be larger, but the quality will be baseline for the camera's internal recording.

[Jody Arnott]

I often use 60mbit for interviews because there is very little movement in the frame, and the codec handles it well. For anything with a lot of detail and movement, definitely 100mbit.

There's no science behind this.. just my own experimentation and observation.

[Gary Huff]

Originally Posted by Jody Arnott

There's no science behind this.. just my own experimentation and observation.

Since it's all subjective (and dependant on circumstances), this is the best way to decide, to try it and see if there is a noticible enough of a tradeoff between image quality and file size.