F55: Looking for 2K 240fps raw MXF to look at

Nice footage Doug.

It's not bad, but look at any edges where the subject is tac sharp in focus and there is a small amount of aliasing. (girl swinging bat, horseback riders right leg,... ) It's less prominent looking at re-encoded footage versus the original, and slightly diminishes itself again after uploading to YouTube or Vimeo.

Maybe nothing for us all to get our knickers in a knot over, but trust me once you put an OLPF on the camera and lower your lines of resolution the difference will be night and day.

I will opt to shoot without the filter hands down.
 
Dennis, I pretty mcuh agree with your assessment of the footage. But the stuff that looked the worst was the bike race. Bright sun, lots of detail, and reallly fast-moving action make it very hard for the camera to process. I'm going to try and get a demo of the OLPF before I pass judgement on whether it is worth it or not, but I suspect I won't want to use it either.
 
Let's be clear about something too, when we are saying "the worst" I think reality is its not even close to being "bad". We're doing some serious pixel peeping here because there was some concern about the line-skipping method being used to deliver HFR HD from 4k full frame scan.

Again, really, its not that bad at all.

Glad you shot and posted this, its testament what the cameras can deliver.
 
Ah . . . but you're not seeing everything that I shot. Believe me, a lot of the bike race stuff that ended up on the cutting room floor didn't look very good, and even some of the stuff that I chose to put in the Vimeo video would not meet my standards for a real production. If you compare "normal" video from the camera vs. 2K HFR, I'd say there is a pretty noticable difference. I'm actually looking forward to the XAVC 180 fps mode in December and may not use the 2K mode anymore if it looks better. Another round of testing will be needed. 240 is nice, but I'll take 180 and better image quality if it comes down to that. And 2K vs. HD means nothing to me anyway, because 2K is just HD with a few more pixels on the sides that are going to get cropped off in the edit.
 
Doug Jensen said:
Ah . . . but you're not seeing everything that I shot. Believe me, a lot of the bike race stuff that ended up on the cutting room floor didn't look very good, and even some of the stuff that I chose to put in the Vimeo video would not meet my standards for a real production. If you compare "normal" video from the camera vs. 2K HFR, I'd say there is a pretty noticable difference. I'm actually looking forward to the XAVC 180 fps mode in December and may not use the 2K mode anymore if it looks better. Another round of testing will be needed. 240 is nice, but I'll take 180 and better image quality if it comes down to that. And 2K vs. HD means nothing to me anyway, because 2K is just HD with a few more pixels on the sides that are going to get cropped off in the edit.
Click to expand...

Like I've said earlier, it will boil down to trading off between aliasing vs softening of images with the OLPF. (extend of aliasing determined greatly by the nature of the scene) Everyone will have their own opinion on it. There are guys on the Sony forums who claim to see absolutely no issues at all :huh:

I don't have any immediate need for 120 HFR at this point in time and can make do with 60p for most of what I need, but its truly a shame the new F's won't live up to the HFR quality of its counterparts... the EPIC and Alexa.



Doug Jensen said:
Are you referring to "HFR Center Scan Mode"? That is coming in Ver 3.0 in December.
Click to expand...



As for center crop HFR (just like Scarlet had) it is coming but remember at the expense of doubling your focal length. Also until we see it there's no guarantee what it will look like or how Sony might derive it. We know how we would design it, but in the end its up to Sony.



morgan_moore said:
Well that should be crisp and free of stepping should it not?

S
Click to expand...


I will no longer believe anything in advance until I see it. Lesson learned now on two Sony CineAlta cameras, there will be no third Sony in my future.
 
Last edited:
morgan_moore said:
Well that should be crisp and free of stepping should it not?
Click to expand...
Well, there's the theoretical, and then there's the actual implementation, we'll have to wait to see how it's actually implemented, but in theory it should be entirely free of stair stepping. "Crisp" is another question entirely; if they implement it as a 2048-pixel-wide direct scan out of the sensor then it should provide for decent sharpness but not on par with what a three-chip 1080p camera could do. It'll probably look somewhere between a good solid 720p and a good solid 1080p, closer to but not quite approaching 1080p.
 
morgan_moore said:
Yep as crisp as the 4k is crisp - which aint true 4k crisp but is still pretty :)
Click to expand...
Only way to get that, would be to have a system that reads the entire 4K sensor at the full high speed frame rate. If that's your requirement, the FS/F series aren't your camera, you need to look at the Epic or a Phantom or something like that.
 
So I've been chatting with Sony about HFR and was corrected that neither the F5 or F55 employ "line skipping" but rather "pixel grouping".

Then it dawned on me.

Binning a group of pixels are what's needed to improve sensitivity or lower noise on a sensor which is exactly what you would need when shooting high frame rates to still maintain some level of noise free image quality. If you didn't do this, you'd need a lot more gain on the sensor read when working in high frame rates because at such read rates the light is reduced.

So its kind of a trade off to group pixels like this, ie, get high frame rates, clean images - but all at the expense of resolution.

So because HFR is coming for center scan mode in December, it has me wondering if Sony will also use this method of grouping like it did for full scan. Perhaps its not so much an issue of sensor high speed read rates as it is the image would be FAR too noisy otherwise.

Food for thought. I'm not getting my hopes up. This does not look good.
 
Dennis,

Pixel binning makes perfect sense when you consider the sensitivity, also there are 11mpix on the sensor, so there are extra pixels as well which we don't know 100% what the purpose is. However this is a bayer sensor too, so how does binning affect the reconstruction of RGB?

And you're right there, just taking the centre portion looses the extra sensitivity that binning would get you.

Ignoring the extra pixel in each group, you'd normally have RG GB, in a binned pixel that would be 4 reds, 4 blues and 8 greens, but how would you best combine those to give you RGB at 2K. The luma is nice and sensitive but the colour components seem more complicated by comparison. I guess taking those 4 reds and 4 blues and averaging them is the quick and dirty way of doing it. But the crux of this is the intelligence of how these pixels are binned and then conversely how they are debayered, using the knowledge that they are binned. This is probably what's leading to aliasing especially in colour channels.

Calling 2K RAW is a bit of artistic license as well, but i digress...

cheers
Paul
 
You must log in or register to reply here.
Back
Top