Sony cameras and the "video look".

maarek said:
Me neither. There has been no need to do anything different in compositing, no matter what it was shot on motion wise. 35mm film, red epic, alexa and the fs100 all match about the same. The only real difference is the rolling shutter that has to be accounted for when doing motion tracking. If you speak with 3d-guys or visit 3d-forums, not one of them has claimed these differences.

There is no magical "motion cadence difference" when all the settings are matched.
Click to expand...

Exactly, the people who claim film has a feel of its own quickly ignore we've been compositing complete digital assets into film plates seamlessly for years. Which is a point I made some pages back. Film doesn't have a unique look that cannot be simulated but what we are trying to do is precisely account for very slight changes from one format to another which some people feel is desirable. Other's not so much.
 
Nirv said:
That's interesting because the mechanical shutter is used to overcome the rolling shutter and the BMCC is a straight rolling shutter. The feel can't be down to the type of shutter.
Click to expand...

I know, I still can't wrap my head around it, but there must be a reason why the Alexa Studio has also a mechanical shutter.
 
Postmaster said:
I know, I still can't wrap my head around it, but there must be a reason why the Alexa Studio has also a mechanical shutter.
Click to expand...

When a mechanical shutter is used in front of a cmos imager they read out from the chips differently and it gives a global shutter type look i.e. without skew.

I understand that mobile phone manufacturers are implementing tiny mechanical shutters too for the same reason.
 
I doubt they use rotary shutters in mobiles. Probably focal plane shutters like in photo cameras.
BTW, a slit shutter like in dual curtain photo camera shutters is in essence a mechanical rolling shutter. Jello effects are not seen because the slit forms at very high shutter speeds.
 
Postmaster said:
I know, I still can't wrap my head around it, but there must be a reason why the Alexa Studio has also a mechanical shutter.
Click to expand...

Well, a spinning mirror shutter is a way to set up a reflex optical viewfinder, which is also clamored for by top end cinematographers.
 
Last edited:
The difference between CCD and CMOS broadly speaking is that in CMOS chips the pixels can be addressed individually as they have electronics on the pixel itself. CCD was column by column. This is why you get charge leaking in CCD and vertical streaks. I think CCDs basically read out quicker although CMOS has probably caught up now. Rolling shutter was very much a CMOS thing. I had a little panasonic lumix with a CCD in and no rolling shutter whatsoever. Great for slapping on a bike or something.

I presume the mechanical shutter just blanks the exposure to give time to read the sensor whilst all is dark and avoiding any rolling shutter.

Love the idea of a mobile phone with a mechanical shutter, can you imagine the noise? :)

And the point made about comping CG onto film is very valid because i've certainly never made any allowances for motion cadence. Just matching colour, blur and grain and it was always okay (been years since i've worked that way with film tho).

So that begs the question whether heavy CGI films show problems for the folk having issues?

cheers
paul
 
Come on CPC, you're the sort of bloke that seems to know a huge amount about these things, so what is the cause of the cadence difference between the BMCC and the Alexa, FS100 etc etc.

Any idea?
 
I think it's a combination of several things in addition to shutter cadence.

We just shot a long form piece in tandem on Kodak Vision3 and Red Epic (B Camera) that was graded at Company 3. When you watch it on our big monitor, both look beautiful... but different. Even though both were graded by one of the best post houses in the world, you can stop it on any frame, compare side by side and note a difference. You can't put your finger on it but it's different. I'm speaking about watching it on our big HD monitor, not once it hits the web, of course.

When you compare frames side-by-side the best way to explain the difference is like this:
- The video-originated footage is crisp and clear. You can usually look at a frame and explain how it was lit (soft key here, fill on the right, etc.).
- The film footage is clean and organic. Artificial lighting looks more natural here and you have a harder time figuring out the set up.
 
Nirv said:
The only person who has a GH2 which he said exhibits the same motion as the BMCC and a FS100 has refused to shoot test footage.
.
Click to expand...

You have a horrible habit of misquoting me. This is the second time I've pointed this out and I find this strange and disconcerting.

I never said the Gh2 has the same motion at the Magic Cam. I said of all the cameras out there, the ones that I feel get close to film are... (insert those cameras here).

If you're going to quote me, and you can't remember what I said, then ask me or re-read the thread.

paulcurtis said:
And the point made about comping CG onto film is very valid because i've certainly never made any allowances for motion cadence. Just matching colour, blur and grain and it was always okay (been years since i've worked that way with film tho).

So that begs the question whether heavy CGI films show problems for the folk having issues?

cheers
paul
Click to expand...

It's never been an issue for me. However, I've never seen much (if anything but Star wars? Anyone know any other movies?) Sony footage with VFX comped in. I could assume that it's because CGi is just CGi. But, I'm not sure.

I also don't think it has anything to do with contrast, saturation, handheld, anything like that. I only chose that video example because I knew it was a run n'gun shoot, etc.

If there was a way to demonstrate I would be all for it. I feel like anyone with that sharp of an eye would be able to just look at FS100, F3, EX1, so on and so forth and just see it.
 
David G. Smith said:
Well, a spinning mirror shutter is a way to set up a reflex optical viewfinder, which is also clamored for by top end cinematographers.
Click to expand...

I don't know for sure,but this is probably why it has a mechanical shutter. David's right, every old school and working DP that I know wants an Optical VIewfinder.
 
paulcurtis said:
However, and again i ask one thing - you suggest that there are digital cameras that you like the motion from - can you elaborate on what they are. The BMCC was one that was cited, although tough to do a side by side with that right now! How about an Epic, what's the motion like on that as in the future i may be able to test one.

I know we've all got real work to do, but as i mentioned above this has become an interesting curiosity for me that may at some point in the future come and bite me.

cheers
paul
Click to expand...

I don't think it'll bite you, at all.

As far as the cameras go (not in any order):

Magic Cam (BMC) - looks almost perfect.
Alexa - looks almost perfect
Gh2 (hacked and at 1/40)
Epic (Or RED MX) -- although sometimes it's in and out for me.
HVX200 - not perfect but was always close enough for me.


Canon DSLRS look a shade better @ 1/45, not so sure why but to me they do.

That's about it that I can think of.

Edit here: Paul, I think that compression is the most logical explanation. I'm not saying that's the reason why, but it seems to be the one thing that's capable of making such a subtle difference like that in cameras that all are based around similar tech?

Just a thought.
 
Last edited:
Kholi said:
I don't think it'll bite you, at all.

As far as the cameras go (not in any order):

Magic Cam (BMC) - looks almost perfect.
Alexa - looks almost perfect
Gh2 (hacked and at 1/40)
Epic (Or RED MX) -- although sometimes it's in and out for me.
HVX200 - not perfect but was always close enough for me.


Canon DSLRS look a shade better @ 1/45, not so sure why but to me they do.

That's about it that I can think of.
Click to expand...

Was just gonna say I like the motion blur on the hvx 200-170/hmc 150 . agreed with the rest to.
 
Nirv said:
Come on CPC, you're the sort of bloke that seems to know a huge amount about these things, so what is the cause of the cadence difference between the BMCC and the Alexa, FS100 etc etc.

Any idea?
Click to expand...

You are giving me too much credit.

I've already written a bit in this topic: http://www.dvxuser.com/V6/showthrea...k-quot&p=1986145683&viewfull=1#post1986145683

The main difference between various CMOS sensors is in sensor read-out speed. Slower sensor read-out = more pronounced rolling effect. This obviously affects motion in some ways. There are also some minute differences between intraframe and interframe codecs in that current interframe codecs use motion vectors to encode frame-to-frame differences which can slightly affect motion rendering, especially with low bitrates like in the fs100. Kholi mentioned comparing the internal codec and external nanoflash xdcam. But xdcam is interframe too.
There is also pixel binning, which can theoretically make an image appear smoother. But I doubt this has anything to do with the fs100 case.
Manufacturers also like to hide things under the hood because they are often too ugly and not matching "specifications". Theoretically, there could be some sensor infrastructure differences.

Make note of the bit about non-disc rotary shutters in the post I linked. Butterfly shutters rotate twice slower than disc rotary shutters. This means film spends twice the time in partial frame exposure compared to disc shutters. In oder words, butterfly shutters have twice as much partial frame exposure artifacts. (This is comparable to slower electronic rolling shutter but with diagonal swipe instead of vertical.) Technically, you can skew the film image if you have an object moving perpendicular to that diagonal. Obviously, that's much more rare situation than horizontal movement opposed to line-by-line readout as in the CMOS case.

As Kholi specifically mentions slower shutter speeds looking better I'd figure the objectionable video looks too crisp.
One thing that I don't think was mentioned is that film grain can actually work as dither on top of the image and the added texture can further enhance the perception of smooth blur. Clean digital on the other hand will lack this, which is in tune with what danstone is describing with Red Epic/Kodak footage.
 
cpc said:
As Kholi specifically mentions slower shutter speeds looking better I'd figure the objectionable video looks too crisp.
Click to expand...

Hmm. Paul said this earlier as well, but then I'd have to question why I couldn't find an acceptable (to me) shutter angle/speed that worked with the FS100?

One thing that I don't think was mentioned is that film grain can actually work as dither on top of the image and the added texture can further enhance the perception of smooth blur. Clean digital on the other hand will lack this, which is in tune with what danstone is describing with Red Epic/Kodak footage.
Click to expand...

I've heard this as well, haven't tried it though.


CPC, on the topic of compression, isn't ProRes Interframe as well? I wonder if you feed the FS100 [hd-sdi] out to something that doesn't record interframe what differences would you see, if any at all.
 
Prores and DNxHD are intraframe.
I don't know if you will see any difference in motion but this would be an interesting test and can filter out the possibility of codec interference.
 
cpc said:
Prores and DNxHD are intraframe.
I don't know if you will see any difference in motion but this would be an interesting test and can filter out the possibility of codec interference.
Click to expand...

Ohhh I thought ProRes interframe, thanks much. That may be something worth doing... I'll have to see which recorder that I have access to can pull a good Pro Res.
 
cpc said:
I don't know if you will see any difference in motion but this would be an interesting test and can filter out the possibility of codec interference.
Click to expand...

Probably a good idea to try different shutter speeds with the codecs as well, since that seems to also have an effect.

Codec would go a long way into explaining the Blackmagic Jesus-cam motion, as I don't really know what they could possibly be doing with the shutter to make the motion look different than other cams.
 
CPC... now that Gary mentions it, let me ask y'something.

The GH2 hack effort has a few different kinds of patches, some are GOP1, some are GOP3. The reason being is that the forum was split on which one looked better for motion, and most (I think) gravitated toward GOP1 patches. So, obviously, this has been a huge discussion there.

My question is this: would Shutter Angle/Speed exhibit a different look depending on GOP settings?
 
You must log in or register to reply here.
Back
Top