AF100 noise

[Rick Burnett]

@grimepoch
Sorry man ..I disagree with you so much when you say things such as "Do you really think that every theatre, every home television is going to be ANYWHERE near calibrated? "
It is absolutely impossible for any Colorist (or a Cinematographer) to compensate to "the lowest common denominator" . eg grandmas 1982 color telly
Images should always be CC / graded , using known specs on a properly calibrated reference monitor.
Its fundamental to the workflow and its not rocket science.
Black is black...white is white and the grey scale is everything in between. A properly calibrated monitor will display these precisely.

(No e at the end, think Grim Epoch)
It's not about always hitting the lowest common denominator, it's about getting the best grade that works the best on the largest distribution of viewable devices within your target audience. The number of people with properly calibrated displays is going to be in the extreme minority in most cases. I do understand the concept of working on a calibrated system, as I said, in the audio world, you do just that. However, before you finalize your master, it is common practice to listen through a set of full range low fidelity speakers to make sure everything is okay. Many studios use Yamaha NS-10s our Auralex speakers (also called Horrorlex in jest) to do this. It's not setting the audio for the worst possible condition, but it is setting it for a very common horrible speaker situation, which is full range, and is a good tool. With a perfectly calibrated display, you've setup your footage to work well in ONCE circumstance. With using a second reference, you've set that material to work in a RANGE of circumstances. I definitely prefer the second.

I take my material off my computer and watch it on a consumer system at the end of my work as well to have a look at what it looks like NOT on my calibrated LCD screen. Not to mention, some people refuse to use LCD screens because they've not found a better replacement than their tube screen for coloring, which is an entirely different thing altogether. This allows me to get a second opinion to what I did with a different signal path.

People are trying to get to the bottom of why the results from different AF100 clips vary so drastically and what is the difference between them. When I film with my AF100, I want the noise to be no greater than what I've experienced with my 7D, which has been FANTASIC in that regard, or at least very very close. From many of the clips, that is NOT the case. With some of the clips, that IS the case. Hence the confusion and question. But further, what has been uncovered is something between capture and Vimeo has caused the noise to be greater than other cameras on the same monitors. Once we can figure out why, and get those values to what we want, then the issue is solved. Whether this be in camera, or in post, or in settings, I really don't care. I just want my footage to look good on the majority of situations and NOT just on a calibrated display.

 

[TimurCivan]

Here is one thing you can do. All of you. I did this, and this is why i get good results.

For a weekend, pickup Friday afternoon with a return Monday morning, so you get the gear for a few days; rent the best monitor you possibly can, a 2k Tungsten light and a 575w HMI. When i first bought the HVX200 i rented a BTLH1700a 2k,575; and sat there with my camera hooked up and went though every single setting and white balance on the camera.

Im lucky in that i currently own a 26" btLH2600, but i still went and got a 2k, and borrowed a 800w joker from a friend, and sat there, put in the work, and tested, tested, tested. Tungsten light, Daylight, high detail, low detail, over exposure, under exposure, different gamma and knee settings etc...

you gotta go through it all and record what works best in what situation.

For example:

This was from Jack Daniel Stanley's Cold Calls, shot with a HPX170 with a Letus Ultimate, and a 85mm L F1.2 @ F2. Using a 500w softbox and three sheets of CTB. Thats three stops down on a diffused 500W though a 35 adapter. I got a good exposure, and a clean image from something that exposed around 25IRE on the waveform. Why? because i knew how to setup the camera to offer noise free performance, in combination with adjusting my lighting to accentuate the strengths of the camera system.

Its my job to know how the camera inside out and backwards.

 

[Rick Burnett]

Does that mean the noise is there? Yes it's there. But it's invisible. Should Panasonic just crush everything in the below-7.5 IRE range to solid black? Maybe. Maybe that's what folks want? If you want noiseless footage, then consider crushing 0-8 ire to black. Maybe that's the post workflow people should employ then, if they're worried about people seeing noise on non-calibrated gamma-stretched displays. Because 0 to 7.5 is all supposed to be black anyway, according to the TV engineers.

Maybe that's a concession people will have to make in this new age when video footage designed for video monitors, is instead going to be displayed on computer monitors... I don't know. It's up to you and how you want to show it.

I don't understand why people have to get so over the top with their examples, like correcting the hulk to magenta, obviously that is an extreme case where I doubt you'd have many circumstances that would be fixing something compared to breaking something else.

Why is it so hard to understand that you can actually grade footage to where it looks good under a range of watching situations and not just to a specific setting? Other professional footage, like trailers for instance, don't have this noise on my uncalibrated display. Obviously they did something to ensure that isn't a problem, and I am talking about stuff that is shot on digital as well.

But let's take this is step further. I am looking at this on a 1 year old mac book pro with default settings. I would consider this a common viewing mechanism for video these days. I think it's fair to say that most people are not going to be calibrating their Macbooks who just like to watch things on the internet, or movies when they travel, etc.

Until I get my camera, I really cannot say. One thing I will say is that I've noticed a lot more detail in the noise in the blacks then other cameras, so I still feel it is a question of curves. And I think your suggestion on things to do in post Barry might be what some people will do to get that footage to something that they feel looks good in multiple watching environments. Since I've not done that specifically to any of the footage, I cannot say one way or another.

I'm glad the OP started this discussion because, regardless if you think it is a problem or not, there are those of us who have purchased the camera that want to minimize that noise in whatever way makes sense in the footage to maximize the viewing pleasure of our material. I think through a lot of experimentation, many of us will be able to provide optimal settings to get the most out of what are targets are.

And maybe I should say, this is not a color correction issue in reality, it's a noise suppression issue.

 

[bgundu]

Do you really think the noise you might be seeing is unacceptable? I think this camera has a lot more wins than losses. Personally, Panasonic have always had the best colors and the magic mojo compared to other manufacturers. Perhaps at the cost of a slightly noisier image. It has never been a hindrance in telling your story. Should I have been appalled that "black swan" was too grainy being shot on 16mm? This argument is getting out of hand. I suspect once this camera get in peoples hands, this issue will die down so we can get back to why we ever picked up a camera in the first place. I know everyone is anxious about getting the camera, so instead of wondering about what it can and can't do, lets focus on what people intend to do with it.

 

[Billygoat]

When shown on an uncalibrated display, looks noisy.

When I calibrated the display, it looks magical.

Which is why footage shot on the really high-end stuff will look pretty magical in the real world of uncalibrated monitors, and especially, projection equipment which tends to have black levels all over the map. It doesn't take perfect black level adjustment to look really clean, because it started that way. But, except for the model/studio shoot, everything from the AF100 has looked pretty clean.

But I am willing to accept two realities:

1. I am buying a piece of equipment essentially sight unseen, which has not even been released yet, and have chosen to on the body of evidence so far, which is that when properly setup, the camera has been capable of cleaner images than the studio shoot.

2. It's a $4700 camera, not a $13,500, $20K, or $40K camera. That may (or may not) require some tradeoffs. But based on what I've seen, since I'm willing to buy the camera before its release, it's a risk I'm willing to take, and there's no need to worry further about noise until the camera is in my hands.

 

[Barry_Green]

I think the issue, if there is one, is that Panasonic has swept the noise under the rug (meaning, under the 7.5 IRE "black" level). And in a properly-set-up video-viewing environment, that works perfectly fine.

But, again, there are folks out there shining flashlights into that area, their uncalibrated monitors are turning the blacks into gray, and they're seeing scads of noise. It's a case of "doctor, it hurts when I do this" followed by the doctor saying "well, don't do that."

If your stance is "people are going to shine a flashlight in there no matter what, and I don't want them to see that noise there", then I would guess that the only solution is to crush that region all down to black then. Which really isn't that much of a big deal, as the 0-7.5 IRE range is all supposed to be viewed as "black" anyway. I don't know what else to tell you.

It's not a totally noiseless camera. But when displayed on a proper-gamut monitor, it does look extremely clean. If you want it to look extremely clean in all possible situations, you'll probably have to scrub out that 0-7.5 IRE range with a noise filter or by compressing it.

But it's not a fault or a flaw. Anything below 7.5 is supposed to be invisible anyway. It's like complaining that your headphones can't reproduce lower than 20 Hz, or complaining that your sound playback device has all sorts of distortion in the range of 0 to 20 Hz. Fact of the matter is, 20Hz is the bottom limit of human hearing, and anything below that is inaudible, so I don't really care what's below there, as it (for practical playback terms) doesn't exist. And if someone goes shifting the phase and somehow magnifying that bottom end to make it audible, then -- they're doing something they shouldn't be doing. And, if you want to prevent those folks, well, you could always just do a high-pass filter and chop off anything below 20Hz, right? It won't have any effect whatsoever on the listener's experience, because they couldn't hear it down that low anyway.

Same thing here. 7.5 is the cutoff point for black. Anything below that is mush and should probably all just be lumped together anyway.

 

[Shooter]

"Does that mean the noise is there? Yes it's there. But it's invisible."


Actually Barry... it is likely the noise is generated by the un calibrated monitors.


Ironic aside: Not long ago that everyone was applying "grain effects" to their videos to make them look like film.
You just got to laugh and hold your breath sometimes Barry!

 

[Justyn]

One thing we can all safely say is that this camera will never be as noisy as the original HVX and that sold like hotcakes and changed the industry forever. I bought it sight unseen and wasn't initially happy with the noise, but once I figured out how to use it and didn't shoot piss poor lighting it was all good. I delivered content for broadcast and a doc that screened at Cannes and never heard a peep about noise or such....

When it's a lot of money to people then there's hesitation and resistance for sure, but if we all had a million for equipment expenditure then 5k would be almost a crash camera. It's all relative..... To a lot of people 5k is outrageous and I just worked with a friend who still shoots with the DVX for all of his content.... It's not one level playing field and we should be happy that there's choices and options....

 

[TimurCivan]

Which is why footage shot on the really high-end stuff will look pretty magical in the real world of uncalibrated monitors, and especially, projection equipment which tends to have black levels all over the map. It doesn't take perfect black level adjustment to look really clean, because it started that way. But, except for the model/studio shoot, everything from the AF100 has looked pretty clean.

But I am willing to accept two realities:

1. I am buying a piece of equipment essentially sight unseen, which has not even been released yet, and have chosen to on the body of evidence so far, which is that when properly setup, the camera has been capable of cleaner images than the studio shoot.

2. It's a $4700 camera, not a $13,500, $20K, or $40K camera. That may (or may not) require some tradeoffs. But based on what I've seen, since I'm willing to buy the camera before its release, it's a risk I'm willing to take, and there's no need to worry further about noise until the camera is in my hands.

Also you should take into consideration that a 5,000$ camera CANNOT be perfect. Look into history. ITs not possible. A chevy malibu is not a Mercedes Benz, you can pretend that it should be, and create two realities.

1. A malibu is made of metal and glass, like a Mercedes, and therefore should perform the same.

2. (reality) The malibu a 20,000$ car, and the Mercedes is a $80,000, there fore the Chevrolet corporation did everything it could to offer as many features as they could for the price point, even if that means a slower 0-60, more road noise and a lack of leather seats.. ( which i think the malibu actually has)

Dude. If the high end performance is that important to you, rent an Alexa for the job.

 

[Shooter]

Grimepoch

"Why is it so hard to understand that you can actually grade footage to where it looks good under a range of watching situations and not just to a specific setting?"


How exactly would you propose that a Colorist or cinematographer set up to achieve this if we dont use the baselines provided?

Black could be the new grey and red the new magenta maybe.

The baselines start in the photography (ligfhting and exposure) and follow thru to the grade and CC. It is set in the physics of light (and electronics)

 

[Rick Burnett]

The only reason I am in this thread is because yes, some of the noise I saw was unacceptable to me. I've seen footage shot with the 7D that had noise that was unacceptable as well, especially anything over ISO1600 usually has these issues. Again, posting about how important your story is really has no relevance to this discussion. If you don't care about the image quality, why not go shoot on the Fisher Price PXL-2000 instead of whatever you use? Obviously there are limits to what all of us can live with. This is a technical discussion.

What worried people was this noise in ISO400 or ISO800, and who wouldn't want to get to the bottom of it if they've just shelled out under $5k for a camera. I certainly would. And I'd rather talk about it now and get a jump on solutions or with information from Panasonic as to what to do or what caused the situation so I can get going faster on my projects. I want to know why my 7D used under similar circumstances has no where near the level of noise in the black. I want to know what processing is different, so I can do things to reduce the noise in those levels.

But, as I said, this hasn't changed my opinion of buying the camera, I can't wait for it to get here.

It's really disheartening that people cannot express opinions on here anymore about the quality of results about a camera anymore. As soon as someone tries to get to the bottom of any issue they might see, the masses jump on top of them. "How dare you criticize any aspect of this camera." I shoot stuff that is for the web only at times, so maybe we just have different requirements in our final results. It doesn't mean I am wrong for having a concern, and that you are wrong for not having one. If this isn't a problem for you, move along, this thread is not for you.

 

[S. Matthews]

I think peoples main issue is not with the level of noise in the af100, but the difference in the level of noise from Leo's shoot compared to others, and weather it was user error, pre production models, workflow, seeing as they didn't have to calibrate to view all the previous af100 footage so why is it such a big difference with the Leo footage.

I had issues with noise as well with the footage but not a huge amount, then when I messed with my calibration a whole lot pooped up (in an improperly calibrated setting) the funny thing was the downloaded version kept exhibiting little bits of grain after calibration (perfectly acceptable amounts) though when I went back to view it on vimeo again the blacks looked perfect and the noise was almost non existent.

I can understand some people's frustration as it was just one video that had these issues, and I can see Barry's and others frustrations as well. though I believe these are not attacks on the camera but more people trying to get to the bottom of what was the cause of that anomaly between Leo's shoot and every other shoot, was it work flow, camera settings, the pre production camera, or was it a simple case of needing black shading, or is it some freak anomaly that popped up for some of us and not others.

either way I think it is an issue that could do with some looking into, if only to squash people's worries.

have a good one and play nice.

 

[Billygoat]

Also you should take into consideration that a 5,000$ camera CANNOT be perfect. Look into history. ITs not possible. A chevy malibu is not a Mercedes Benz, you can pretend that it should be, and create two realities.

1. A malibu is made of metal and glass, like a Mercedes, and therefore should perform the same.

2. (reality) The malibu a 20,000$ car, and the Mercedes is a $80,000, there fore the Chevrolet corporation did everything it could to offer as many features as they could for the price point, even if that means a slower 0-60, more road noise and a lack of leather seats.. ( which i think the malibu actually has)

Dude. If the high end performance is that important to you, rent an Alexa for the job.

Yep, my point exactly. There's no such thing as too good to be true, and still true, and there's no free lunch. Still this camera is probably an unprecedented value. I would equate it more with my 2000 VR6 Jetta, (now with 163,000 miles), than a Malibu. It's a really outstanding product for the money.

 

[f64manray]

Also you should take into consideration that a 5,000$ camera CANNOT be perfect. Look into history. ITs not possible. A chevy malibu is not a Mercedes Benz, you can pretend that it should be, and create two realities.

1. A malibu is made of metal and glass, like a Mercedes, and therefore should perform the same.

2. (reality) The malibu a 20,000$ car, and the Mercedes is a $80,000, there fore the Chevrolet corporation did everything it could to offer as many features as they could for the price point, even if that means a slower 0-60, more road noise and a lack of leather seats.. ( which i think the malibu actually has)

Dude. If the high end performance is that important to you, rent an Alexa for the job.

Just to keep our analogies straight. The Camry, Accord and the Hyundai Genesis are often compared to the likes of BMW and Mercedez, and are considered to perform conquerably to the higher end models. Especially, I remember when the Hyundai Genesis was released and it was compared to $60-80K german models. But hey, if you want the status, Hyundai Genesis may not cut it even if it is identical or nearly identical performance and build. if you look at performance histories in CR, it's a wonder that BMW and Mercedez are even still in Business. Honda and Toyota out perform them in mechanical and build quality down the line. If you were to give me a car and I didn't have to worry about the cost, I would still choose the Honda Accord V6, manual, Coupe over the German counterpart. It's one fine machine.

You may find some amatuer photographers that are seduced by the cost and build quality of a 1DsIII, but I can point to many more pro studios that operate with a Canon Rebel at the heart of the business. Image output (still) of the T2i is pretty fine and hard to argue with. Not to say that many pros don't shoot the 1D series.

Everytime I hear this "It's a great cam for $5K" it kind of drive me up the wall. It either meets the bar or it doesn't. I DON'T CARE ABOUT PRICE. You can intercut it with 35mm film or you can't. The 5D2 is capable of doing this according to Tattersall, Hurlbut etc. If the AF100 isn't clearing the bar in this regard, maybe I just need to stay with the cheaper 5DII. Just for grins, I'd like to see another Zacuto shootout with the AF100, Arri Alexa, F3, Red MX.

 

[Rick Burnett]

Grimepoch

"Why is it so hard to understand that you can actually grade footage to where it looks good under a range of watching situations and not just to a specific setting?"


How exactly would you propose that a Colorist or cinematographer set up to achieve this if we dont use the baselines provided?

Black could be the new grey and red the new magenta maybe.

The baselines start in the photography (ligfhting and exposure) and follow thru to the grade and CC. It is set in the physics of light (and electronics)

Part of the baseline starts in what exactly the sensor is doing, and what is happening with the curves being applied to the sensor data in camera before it gets knocked down to 8-bit. This is about getting that right first, or, understanding why the noise is more pronounced in that range based on those settings in the camera. Since I am happy with the noise response of my 7D, I will easily be able to take the footage from both and compare and figure out if I can get the noise to be around the same depending on settings.

All sensors are not created equal and all noise is not created equal as well. It just want to tame it some more myself. I would not use any of those noisy clips are they were in anything I released personally (other than test footage) so I want to get to a solution, which may be as Barry has suggested. I think the camera DOES show more information in the blacks which would explain that noise being so prevalent on an uncalibrated monitor, and thus, we might start to get to the reasons why we are seeing differences.

I mean, take it from this perspective, I've seen GH2 ISO3200 footage that had less noise than the AF100 at ISO400 posted. Can you imagine the confusion in seeing such a thing. And I downloaded the GH2 footage as well and from a noise perspective, it was very clean. I have been absolutely amazed at ISO3200 on the GH2. Again, this was on an uncalibrated monitor, but at the same time, it just didn't make sense.

 

[bgundu]

Well I have no issues talking technical, but to complain about the noise on a pre-production camera after Barry and others have already explained their workflows seems like we are beating a dead horse. By all means, once this camera is released, let the games begin. It won't be long before it here

 

[Rick Burnett]

Well I have no issues talking technical, but to complain about the noise on a pre-production camera after Barry and others have already explained their workflows seems like we are beating a dead horse. By all means, once this camera is released, let the games begin. It won't be long before it here

Well at least it's only a couple weeks now! Alright, who is the first person that will be shooting (1) carnival, (2) nightime in a city, (3) slow motion water fall. You know, the typical array of FIRST videos.

 

[heavyG]

Well I have no issues talking technical, but to complain about the noise on a pre-production camera after Barry and others have already explained their workflows seems like we are beating a dead horse. By all means, once this camera is released, let the games begin. It won't be long before it here

Exactly! Let's revisit this in a couple of weeks when we have production cameras in hand and we can test multiple settings under varied conditions… Until then I don't really see the point of talking too much about an initial one off test from a camera that was an engineering prototype. One test out of several that displayed noise only to some who viewed it.

If two weeks from now this is a problem with production cameras… there will be something to get to the bottom of - regardless of the cost of the camera.

 

[heavyG]

Well at least it's only a couple weeks now! Alright, who is the first person that will be shooting (1) carnival, (2) nightime in a city, (3) slow motion water fall. You know, the typical array of FIRST videos.

OOOH! Can I do slow mo pouring liquids?

 

[f64manray]

Well at least it's only a couple weeks now! Alright, who is the first person that will be shooting (1) carnival, (2) nightime in a city, (3) slow motion water fall. You know, the typical array of FIRST videos.

Oooh! I can do over cranked sleeping cat.