The Footage from Leo's shoot...

Status
Not open for further replies.
First off good stuff Leo and thanks for posting.

Im sorry are people watching these vimeo clips in full screen on a computer screen?

I have an apple 30" Cinema display that I use for my user interface and watching ANY of the AF100 clips posted on vimeo or ANY of the F3 clips posted on vimeo at full screen - at this screens resolution (and yes this has been RGB calibrated) I can pick any of them and pull a shot up that is seemingly full of noise. In fact as a test I just went through a couple of the F3 clips and saw some horrible blocking, but I am also smart enough to realise a/ its been heavily compressed b/ I am then watching that heavily compressed clip stretched from 1080 to almost 3K c/this screen isnt and cannot be correctly calibrated in its current configuration and d/ see a again.

But hey I dont use this screen as a reference monitor - nor do I use it to judge anything. If I want to actually view the noise level of a camera - I'll personally shoot it or be there when someone else does and view the result on my reference monitors - a JVC DTV CRT monitor via HD-SDI or my Panasonic Pro Plasma via HD-SDI with a REC.709 compliant signal.

I can point out the things I saw on the F3 stuff I just viewed, but I wont. It's a pointless exercise of the ridiculous.

So here's my experience with the AF100. I spent all day with it last week. I wasnt able to produce ANY aliasing. I was able to produce a ridiculously small amount of moire by basically doing everything in my power to make it happen and even when I did it was such a small area that at 6 feet from my pro plasma I couldnt see it.

I shot a bunch of stuff in the studio, dropped it on a calibrated broadcast monitor. The guy standing next to me when I did was a cinematography lecturer at one of the largest film schools here. He said "oh nice". The other guy was Matthew B from this forum (former student of the aforementioned), and I'll let him tell you in his own words what he repeated to me all day. I do remember clearly him saying at one point "Im just not seeing anything that people are worrying about". Needless to say he's got one on order.

I really stretched the camera out in some shots in ways that you just wouldnt shoot - my point was to pull it back to where I liked it the most. And still only came across one shot I wasnt happy with - that was a shot with DRS on the highest setting - no noise in the image - just a bit of ghosting? I guess you'd call it.

There is some noise in the AF100, but not by anyones standard more than I have seen in any production camera, I'd say it rates rather highly in the scheme of things.

For the record I shot some totally beautiful shots on the 5DMKii recently, but how I wished for natural looking water.

Heres a shot from the AF100 with many gradations. I thought Id post this as a still as it may suit computer monitors better. Shot on a heavily overcast day sun coming through the clouds camera right. Compressed JPEG from original AVCHD recorded file.
 

Attachments

  • bridge.jpg
    bridge.jpg
    58.4 KB · Views: 0
Thanks for posting that image. This is EXACTLY what I am expecting. VERY clean indeed. Thanks for your perspective.
 
Footage looks great (actually, I admit I'm more appreciative of other people's talents, more than my own). Makes me want to learn more about lighting. I can see great things being done under the right conditions. You guys knocked me out of the ball park. I would really like to see a scene or two shot with people with a good sense of story, direction, visuals, and acting and utilizing this camera. This camera is capable of far more than just shallow depth of field and low lighting.
 
Well here is my 7D noise test. The lighting isn't exactly the same... or as good... since I only used 3-5 candles, and set it up very quickly, but the lighting isn't exactly the point... the point is the noise. Which IMHO, on my custom settings, my 7D at ISO 1250 looks like the AF at ISO 400. It may not always be this case, it could be over-compression on the AF footage, but I seriously doubt it because the first shots with the fruit looked very crisp and natural. My workflow was H.264 to proress for Final Cut and then back to H.264 with compressor for vimeo. So the same compression that people are blaming the noise on should have also been introduced in my footage as well. Either way, I hope it's helpful and helps people reach a verdict either way... Since now, no matter what your monitor is calibrated to, one or the other should clearly be showing more noise than the other.

It should also be noted that afterwards I did some color work and was able to bring up the shadows even further with no noise. I'll get around to posting that version later tomorrow.

http://www.vimeo.com/17356308

And yes... the 1080p version is available for download.
 
I've watched the footage about a thousand times and it's making me and the crew happier then a little girl getting a pony for her birthday. But seriously this is great footage and I'm glad Barry chose the right crew to give it to us. Thanks a lot for this guys!

Cheers!
 
mrmoonay said:
Well here is my 7D noise test. The lighting isn't exactly the same... or as good... since I only used 3-5 candles, and set it up very quickly, but the lighting isn't exactly the point... the point is the noise. Which IMHO, on my custom settings, my 7D at ISO 1250 looks like the AF at ISO 400. It may not always be this case, it could be over-compression on the AF footage, but I seriously doubt it because the first shots with the fruit looked very crisp and natural. My workflow was H.264 to proress for Final Cut and then back to H.264 with compressor for vimeo. So the same compression that people are blaming the noise on should have also been introduced in my footage as well. Either way, I hope it's helpful and helps people reach a verdict either way... Since now, no matter what your monitor is calibrated to, one or the other should clearly be showing more noise than the other.

It should also be noted that afterwards I did some color work and was able to bring up the shadows even further with no noise. I'll get around to posting that version later tomorrow.

http://www.vimeo.com/17356308

And yes... the 1080p version is available for download.
Click to expand...

I can confirm that I see less noise in this particular video than in the AF100 video we're discussing in this thread, both viewed on the same displays. However, noise aside, the overall picture quality of the AF100 video blows this 7D footage away. There's no question about it, to me at least. I've seen A LOT of projects shot with DSLRs (5Dmk2, 7D, and GH1/GH13 mostly), and have worked with all in post, and I have never seen anything from any of those compare to what Leo, Barry, and co. shot for this video. Noise or no noise isn't the end all be all. It would be nice if it wasn't there, but if you're aware of it and what causes it to be pronounced you can shoot to avoid it or try to reduce it in post if you see the need. It sometimes can actually help the aesthetic too.
 
I agree that the AF100 image seems to be bounds and leaps better.

Particularly in resolution there's seems to me to be no contest between the AF100 and any DSLR in that matter, which was expected.

However (again, I might be wrong) the noise seems to be pretty pronounced, even at ISO 400 and on brightly lit scenes.

This might be due to the workflow, etc, but Panasonic does have a history with noisy cameras, so I wouldn't be surprised if this turns out to be the case.

Other point is DR. It doesn't feel like there's 10 1/2 stops in any of those images. Again, I might be mistaken but it's my general feeling.
 
Last edited:
Mike@AF said:
I can confirm that I see less noise in this particular video than in the AF100 video we're discussing in this thread, both viewed on the same displays. However, noise aside, the overall picture quality of the AF100 video blows this 7D footage away. There's no question about it, to me at least.
Click to expand...

Really? I'd say the AF-100 is an improvement over 7D/5D footage... but not that necessarily it blows it away... especially in DR and highlight areas. But I think all of these large sensor cameras and DSLRs have reached a quality threshold where it's going to be hard to "blow" one another away... Unless were speaking of Red's 5k RAW with HDRX viewed on an imax screen or something like that. The shots of the models with the AF-100 still looked pretty soft to me. Which isn't bad, because an image that's too sharp looks more digital and video like. Either way, thanks for the input. Hopefully we'll get closet to pinpointing the cause of the noise. Even though I really do suspect it's just that Canon has better noise reduction in general... and that allot of the noise in the 7D is lost in the down-sampling from the 18MP sensor block.
 
Last edited:
What I saw this weekend was very clean and the things I pointed the camera at were not properly lit. Any computer LCD screen will display higher levels of noise and taking vimeo compression in account I don't see any need for concern with regards to noise.
 
Have you been to beirut? 56k

ilauzirika said:
downloading it right away!!!

Edit: ouch! 3 hours until it is downloaded......damn 1mb internet connection.....those are the wonders of living in what is considered a rural area.....500m down the road they have nice cable internet, not here......
Click to expand...
 
Stupid me. I didn't realize the shot after the greenscreen shot was the keyed version of that shot until now. HA! Very good key/greenscreen for 4:2:0. Properly lit I don't see what the big deal is about not having 4:2:2.

Also, as I watch this video over and over again the noise stands out less and less. It actually doesn't really bother me anymore. I just wanted to figure out what was causing it. And I don't think it looks soft either. As sharp as it needs to be (you don't want too sharp) and very filmic video, in my opinion.
 
LOL I go away for a day and the whole internet has gone from screeching that the AF100 sucks because its highlights clip worse than anything in the world ever, spend the day actually getting a bit of hands-on with the camera and seeing some footage shot by an experienced cameraman. Come back and discover that the internet is now screeching about the noise in AF100 footage.

I've seen the monitoring signal from the camera in a fairly dim studio (lit with a couple of kinoflows) and it looked absolutely lovely. It had a roomful of experienced owner/producers-directors-cameramen who make their entire living from the cameras they own reaching for their credit cards.

The example someone posted of two frame grabs with different players showing the footage with different gamma/blacks tells the story I think- it is a mismatch issue between RGB and video black levels or gammas. It possibly means those of you seeing awful noise have got some different settings or haven't got a certain colour profile loaded or something like that. I saw absolutely not a trace of noise at ISO800 from the camera output live to a large TV screen in the studio. The footage was sumptuous and gorgeous. It is purely a question of the right display curves being loaded up to look at the footage.

Hywel.
 
Joe Walker said:
Wow!!!!! Okay kids listen up! I just drove all the way to a friends house and watched the vimeo clip on his Dell 23" monitor and holy crap, yes it had noise! On my Apple 30" it looks spectacular but over here on this Dell 23" it looks grainy, it's like this thing has a whole different gamma curve or something. The plot is thickening!
Click to expand...

Hey Joe, thanks for checking it out.

I mentioned noise early in this thread but never found it so distracting like others.
So now I got interested and tested:
1. watching the vimeo-compression with Mac-OSX QuickLook on my Dell 24": haha, the image falls apart and looks like my earliest phone from a Siemens phone 2006.
2. Watching it on the same screen in Quicktime...ahhhh now we´re talking. There is some noise on the 800ISO shot and loss of detail at her skin there.
3. Convert to ProResHQ and put in timeline and watch via MXO2 on my production screen.....getting even better. That screen is making a EX-1 look noisy and the AF-100 matches that :P
4. Watch it on my plasma TV (Panasonic brand of course): wow.....the consumer TV adds contrast which takes away all noise that might have been there. Images are clean and have a great falloff in colour and detail. So a little post-correction would help the computer-watching also.

Now to be fair: the F3 stuff looks cleaner on all these screens also in the RAW film-version which is also on vimeo. But hey, that was recorded via 4:4:4-HDCAM-SR. Also some of the Osaka stuff was cleaner than this. so the camera is capable.

So I have two suspicions: 1. was DRS on that pushed the blacks a little more than needed? Because if I see noise in Quciklook tgere is probably something there that´s not in the Osaka stuff.
2. how exactly did you convert in Compressor? Apple Preset or your own setting? I usually use the youtube-sharing preset by Apple and that works fine.

Overall there is nothing that can match this cam (yet) in this pricerange or even near it. overcranking 1080p alone is worth the price.....
 
Last edited:
mrmoonay said:
The AF-100 sensor only has a 3-4MP, if I remember correctly,
Click to expand...

Well this would be totally incorrect. The effective number of pixels on the AF100 chip set for imaging is 12.4mp. While I cannot explain why some of you are seeing noise in this picture, it looks absolutely fantastic on my computer. And I trust it will look great when I get it to a large screen, with luck today. But I working a show and sometimes resources are not what you would like them to be, so we will see.
 
Subjective assessment over a variety of equipment won't cut it. People will see different things depending on their expectations. Then you're judging expectations, not the camera.

Any one with a scope to do a signal to noise ratio test? Or the simple expediant of putting the lens cap on and shooting the black with various ISO settings then do a difference test to see what the sweet spots are of the ISO?

On the DSLRs the ISOs were not noisy in a linear fashion. No doubt its the same here.

I'm not familiar with Panasonic picture profiles, but with both the Canon and Sony PPs the settings make a lot of difference. In those anything that boosts any level significantly, like DRS, huge color changes, white balancing, etc adds noise. I'm sure the AF100's PPs will too. BTW, DSLRs hide noise by crushing the blacks.
 
Last edited:
I have put a few HVX, EX clips together with this studiofootage on a timeline and watched on my miserable Dell 24", which really shows noisy everywhere. Mind you it is my monitor to edit, not the monitor where I watch what I edit ;-) I just wanted to see what others "see".

The HVX is down another league or maybe even 2 leagues. I still love it outdoors, but it is looking´dreadful on the wrong monitor when in dark spots compared to these other cams. The EX at 6dB is probably where the AF100 is at 400 ISO (my unscientific guess). It is a different kind of noise on the SONY, but it´s there and a lot on the Dell. What they are able to do though is hold up the detail. And the EX is absolutely the league I for my part would like to get in terms of sensitivity and noise.

Af-100 seems to have the EX sensitivity with the 7D DOF and the HVX workflow. Does it get better than that? Now I just need a project to pay for it ....
 
Ok,

Here is a solution: Please render out an UNCOMPRESSED AVI. That would allow for a difinitive method to see what it looks like. I know, it would be a large file, but then, so what?
 
Don't know what all the complaining is about, but how are you guys (the Complainers) viewing the video. I downloaded the 35mb file from vimeo, and played it back through Premiers CS5 and MXO2 onto 1080p LCD and Sony CRT studio monitor and there was hardly any noise to speak about. Just a little is 800iso. This video really looks stunning, can't imagine it looking better than this even with the 386mb file. So maybe before complaining you should really check your monitors to see that they are properly calibrated. Hats off to the crew and cast.
 
I had a look at the ProRes footage from the FTP (thanks!) and am mostly happy with what I saw, given that parts of this clip are torture tests of sorts. I'm aware that ProRes is essentially visually lossless, so this is nearly as good as looking at the AVCHD footage out of the camera. Just to confirm, this hasn't been through multiple lossy compression generations, right?

My honest thoughts:

1. There is aliasing on the first static shot, but to me this looks like the footage was interpreted as interlaced? Maybe the guys who edited it can confirm. The test chart at the top is aliased. If that's not a post issue, that's a little disappointing, but I imagine that designing optical lowpass filtering to make every single conceivable image look non-aliased is an incredibly difficult balancing act.

2. I can see sensor noise in the shadows, sure. It's mostly monochromatic and isn't ugly looking. This is in line with my expectations. EDIT: Windows Quicktime player produces an elevated black level from my experience. This is perhaps why there's so much disagreement on this (non?)issue.

3. The lemon squeeze shot is a festival of compression soup, but the sensor noise + flying lemon particles make this a very spatially complex shot. Almost the entire screen has high entropy moving detail, so that's understandable. If I was shooting exploding fruits, I would definitely want to use an off-board recorder! MPEG-2 went blocky, and AVC goes blurry instead - not sure which is worse!

4. The green screen segment doesn't look very good, if I'm being honest. I realise this is probably just a quick test. What was it keyed with? It looks like whatever plugin was used doesn't use any sort of chroma smoothing hence the pixellated edges.

5. There is some ringing/overshoot on the edges of the champagne flow and on the black/white lines in the test card near the end. I assume that can be minimised in camera. In terms of compression, I wonder if it'll be better to shoot softer and sharpen in post. Being a Windows user (and having access to AviSynth) I'm spoiled for choice when it comes to selective sharpening.

Thanks very much for doing the test and putting the camera under such grueling conditions - and also for sharing the Prores footage. It looks like a great tool.
 
Last edited:
Status
Not open for further replies.
Back
Top