FS5 - R U shooting 25/30p or 50/60p?

[Ron Evans]

Originally Posted by Chad Johnson

50mbps. That's 50mb per second no matter what the frame rate is. So shoving more frames into that second results in each frame having less resolution. Less resolution may not show much if you don't grade your footage, but push it a little and it breaks down. Not to mention you lose the amount of light hitting the sensor. But mainly, it just looks like video tape - all glassy. Nobody likes that look. You may as well shoot at 60i if you are trying to look like video tape.

The math doesn't work that way for LongGOP encoding. Only the differences are recorded most of the time whether the frame rate is 30 or 60. Since the differences will be smaller at 60 and much larger at 30 the difference data for 60 will be less ( but twice as often ) as the bigger change needed at 30 and thus requiring more data to be recorded. Overall the data requirements I think are minimally different. The bigger difference is the processor is working hard to sample twice as fast for 60 compared to at 30. Likely why the FS5 will only do 30P in UHD compared to the FS7 etc.

An advantage of shooting at 60 with 1/60 shutter ( or 50 at 1/50 ) is that on a 60P ( 50P ) timeline the video will be smooth motion of video and on a 30P ( 25P ) timeline will be just the same as if it was shot at 30P ( 25P ). At least that is what happens in EDIUS.

Arguments may change for an iFrame encode but the FS5 is all LongGOP just like the X70 for example and closer to the consumer XAVC-S format but in a MXF wrapper.

Ron Evans

[Christopher Young]

Nigel if you have the time could you possibly run a 200fps clip through MediaInfo and see what it says? Just go to View and select Text and you will have the entire clip structure in front of you

https://mediaarea.net/en/MediaInfo

Only if you have the time as I'm curious.

I just checked the 700 again plus checked the footage on the timeline. The camera is definitely scanning at 200fps but it buffers it out and records it as 25p.

Chris Young
CYV Productions
Sydney

[Anthony Mozora]

Originally Posted by Noa Put

Here Chad, below video is a mixture of 1080p 50fps 28mbs avchd and 50mbs ipb mov and 100mbs 4K 25fps.

since you claim that 50fps looks like video and since there are plenty of shots in there that are either at 25fps and 50fps it should be easy for you to see the difference, no? Also, because math is all saying, could you also pick out a shot which is at 28mbs vs the 50mbs codec?

https://vimeo.com/136085870

NICE work here Noa!

can I have a link for your vimeo ?

[Nigel Davey]

Originally Posted by Christopher Young

Nigel if you have the time could you possibly run a 200fps clip through MediaInfo and see what it says? Just go to View and select Text and you will have the entire clip structure in front of you

https://mediaarea.net/en/MediaInfo

Only if you have the time as I'm curious.

I just checked the 700 again plus checked the footage on the timeline. The camera is definitely scanning at 200fps but it buffers it out and records it as 25p.

Chris Young
CYV Productions
Sydney

Hi Chris I did what you asked with MediaInfo and the frame rate is definitely showing at 50fps for the 50p clip at 200fps. Likewise I also recorded a 25p 200fps clip and it shows as 25fps in MediaInfo.

[Christopher Young]

Thanks for that Nigel. Just confirms my belief that the the overcrank settings although having the same fps available in the FS700 and FS5 are quite different as MediaInfo definitely states 25p for the 700 when running a 200fps.

Chris Young
CYV Productions
Sydney

[John Wiley]

The FS700 can record slow motion as either 25p or 50p (in PAL mode) - it is a menu setting.

[Christopher Young]

John - Really? I missed that one. Please shine the light for me and show me the error of my ways :)

Chris Young
CYV Productions
Sydney

[Christopher Young]

Got me hunting John.

Sorted it. So long since I've used the slo-mo I couldn't even remember where the menus was. Had to go back to the manual... when all else fails... page 52. I see now. If you set it to 25p in the Super Slow Motion menu at 200fps you actually get 12.5% slow and if you set it to 50p you actually get 25% slow motion. So it all depends just how slow you want it I guess. This camera was set to 25p so that's what threw me.

Thanks for pointing that out.

Chris Young
CYV Productions
Sydney.

[Nigel Davey]

Just thought I'd update folks over this. I wrote to Sony Prime Support and asked if the quality of 50P 50Mbps was degraded over 25P at 50Mbps in XAVC HD. I used the logic expressed further back in this thread, ie with a cap of 50Mbps and twice as many frames, 50P should have half the quailty compared to 25P which is getting the full 50Mbps. This is how Sony responded:

"That way of explaining would be correct in an Intra codec. However, since you record in XAVC Long GOP, there is no such "double quality" in a 25p frame compared to a 50p frame"

I'm afraid that's all I got back and I'm not technical enough to know why this is true.

Anyone else?

[Noa Put]

So no quality loss at all in (XAVC Long GOP) 50P compared to 25p? In that case I always would shoot in 50P.

[Ron Evans]

Originally Posted by Nigel Davey

Just thought I'd update folks over this. I wrote to Sony Prime Support and asked if the quality of 50P 50Mbps was degraded over 25P at 50Mbps in XAVC HD. I used the logic expressed further back in this thread, ie with a cap of 50Mbps and twice as many frames, 50P should have half the quailty compared to 25P which is getting the full 50Mbps. This is how Sony responded:

"That way of explaining would be correct in an Intra codec. However, since you record in XAVC Long GOP, there is no such "double quality" in a 25p frame compared to a 50p frame"

I'm afraid that's all I got back and I'm not technical enough to know why this is true.

Anyone else?

As I mentioned in my previous post. If you are recording i Frame then every frame is encoded separately as Sony has said. However for LongGOP each frame is not encoded separately. The first frame is a full encode , just like i frame encoding, but the next frame only records the difference from the first, etc etc Depending on the encoding codec there will be other frames during 1 sec that are also full frame encode or predict backwards from the known data and then the process repeats. The net is that the LongGOP encode is potentially better than the full iFrame encode at substantially lower data rates when one is encoding 25P iFrame and the other 50P Long GOP. The 50P encode has a lot more data points to sample but the difference each time is small hence the data needs are lower. The number of encode sample for 25P in 1 sec is 25 full frame encodes for I frame. For 50P Long GOP the samples are 50 but only differences from the first full frame encode ( simple explanation only ) are needed. If nothing moves there is almost no data recorded. However for Intra Frame recording the whole frame would still be encoded. Since at 50P the movement is small from each frame the data needed to just record the difference each time is also small. For both 25P and 50P the data requirements are down to how much difference information has to be recorded. There is more difference information at 25P ( there is more movement between frames than at 50P ) so data needs are more than at 50P for each frame recorded. I expect that at a data rate of 50Mbps it is a wash.

Like Noa, I would always record 50P ( or 60P in my case ) since you can always get the slow frame rate cadence back if that is what you want, in editing.

Ron Evans

[Nigel Davey]

Thanks Ron, that's useful. So is there any visual difference (after export) of turning 50P into 25P? I assume it just throws away every other frame.

But something I have been wondering about shooting in 50P, is what about the principle of doubling the shutter speed? Shooting at 1/100 will obviously need more light? Or is it not noticable shooting at 1/50 and then converting down to 25P?

[Chad Johnson]

Yes you will lose some light shooting at the proper shutter speed for 50p. So I still say, if you're not going to go to slow motion in post, use the frame-rate of the project you're editing in.

I still don't get how, even in long gop, you're using the same amount of data per frame at 50p vs 25p. When a codec says 50mbps, is that what it's at, or is it "up to" that in the case of long gop?

[Doug Jensen]

Here's an analogy. It's not perfect, but it sort of explains it.

Suppose you had to type 25 pages per second with a typewriter. Every page stands alone and has to be typed from top to bottom from scratch. Now suppose you only have to type 10 pages from scratch per second, and you can copy and paste another 40 pages with only a sentence or two of changes on each one. In the first scenario you are working very hard to create 25 pages, but in the second example, your output is 50 pages but the effort is actually the same. That is the root difference between XAVC-I and XAVC-L even if the numbers in my analogy aren't quite accurate.

[Chad Johnson]

I understand how Long GOP works (sort of) I just don't get how twice as many frames (or changes between frames) within 50mbps can have the same quality. Unless when shooting 25p you're not actually using the full 50mbps.

Anyway, if every frame of 60p footage in XAVC-L is as good of quality as a frame from 30p, that's good enough for me. Glad to know. I still won't just be shooting 60p all the time.