Jump to content

ETTR: Noise, Dynamic Range and Skin Tones


jonpais
 Share

Recommended Posts

1 hour ago, BTM_Pix said:

Is there something in the colour of the lights that was causing the raised levels in the blue channel with Cinelike D perhaps?

That was my initial guess too, but it would be strange for just one photo style to behave like that, no? 

Link to comment
Share on other sites

EOSHD Pro Color 5 for Sony cameras EOSHD Z LOG for Nikon CamerasEOSHD C-LOG and Film Profiles for All Canon DSLRs
  • Super Members
54 minutes ago, jonpais said:

That was my initial guess too, but it would be strange for just one photo style to behave like that, no? 

Doing the same thing with 400% crops from my test earlier day (which was daylight obviously) you do see differing things going on in the blue channel. Not just the overall level (which you can see is higher and in a different range as are the other colours which are more bunched in the mid area) but you can see it gets rolled off at the top.

So there are differences but its certainly looking like the studio lights shift it a lot more noticeably.

Natural profile on top, Cinelike D below it.

 

Screen Shot 2017-06-15 at 20.46.16.jpg

Screen Shot 2017-06-15 at 20.46.37.jpg

Link to comment
Share on other sites

  • Super Members
9 minutes ago, jonpais said:

If it makes any difference, I was using LED lighting set at 5600 K.

I'll do a few tests with a couple of different ones I've got around if I get a chance over the weekend.

They're all varying degrees of cheapo so I expect it'll be all over the place !

Link to comment
Share on other sites

1 hour ago, BTM_Pix said:

I'll do a few tests with a couple of different ones I've got around if I get a chance over the weekend.

They're all varying degrees of cheapo so I expect it'll be all over the place !

I could also set my light to tungsten and do another test. Wish I had a model though! Maybe this afternoon I can get my hands on one (ha). I'm dying to know where you're going with this. I think you are our Thomas Edison or something.

Link to comment
Share on other sites

  • Super Members
5 hours ago, jonpais said:

I'm dying to know where you're going with this. I think you are our Thomas Edison or something.

Haha.

Where am I going with it?

I think you and I are in accord over how much received wisdom there is with much of this stuff. And truth be told I'm as guilty as anyone else for perpetuating much of it!

When you consider the image controls available in a G7 etc you have the following :

 

8 Base colour profiles

4 Individual modifying parameters to those 8 (Saturation, Sharpness, Contrast, Noise Reduction) each of which have 11 possible settings (+5 to -5 and 0)

2 Global parameters (Highlight and Shadow) both of which have 11 possible settings (+5 to -5 and 0)

 

That means that there are over 1.7 million different combinations.

And thats without the additional Hue control that some Lumix cameras have.

Oh and thats 1.7 million different combinations per profile.

Can anyone honestly say they've tested all 14 million combinations to determine their definitive setup for these cameras?

Obviously no one has, even though I know sometimes it feels like we have ;)

And on the basis that all of these parameters do actually make a difference, there could be some really interesting combinations of them that no one tries because the received wisdom (and lets be honest with 14 million combinations its no bad thing to use that) tells us that we should use a smallish core base and then tweak from there.

Do we actually know, for example, if you have the noise reduction at +5 and the shadows at -5 and the highlights at +4 on a Natural profile with the saturation at 3 and the contrast at 2 and sharpness at -3 that these cameras don't suddenly look like you've got Kodachrome 64 in them? No, because we'd guess that those settings would make it look like it horrendous and not go near them. Does anyone even really know what happens if you have the noise reduction at +5? Does anyone even really know if it actually works at all as I can't imagine most people have ever tried it at that setting.

And the reason we don't know, in my opinion, is that not only are there so many combinations but that its just too fiddly to be diving in and out of menus to change them and have the immediate feedback of what the interaction of those parameters actually looks like.

So, where I'm going is to make that a lot easier and hopefully people can come across some happy accidents and make profiles that they don't have to spend hours tweaking later.....

Link to comment
Share on other sites

I like where you're going with this! i don't think it's as complicated as all that though. For picture profile, I'd choose the one with the greatest dynamic range, for sharpening, as sharp as possible in camera without ringing or haloing, and so on. But I still think everyone should do their own testing to see what works best for them, and if you've got an idea to make that easier... here are three more shots with the LED set at tungsten 3200 K, ISO 200, f/3.2. (1) Cinelike D, (2) Standard (3) Natural. Settings: Sharpening -3, NR -5. Even with tungsten, there's a peak in the blue channel in the shadows using Cinelike D, and with the other two profiles, there's a bump in the red channel in the highlights. Histograms (1) Cinelike D (2) Natural (3) Standard

Cinelike D Tungsten (3200K).png

Standard Tungsten (3200K).png

Natural Tungsten (3200 K).png

Cinelike D.png

Natural.png

Standard.png

Link to comment
Share on other sites

  • Super Members
1 hour ago, jonpais said:

I like where you're going with this! i don't think it's as complicated as all that though. For picture profile, I'd choose the one with the greatest dynamic range, for sharpening, as sharp as possible in camera without ringing or haloing, and so on. But I still think everyone should do their own testing to see what works best for them, and if you've got an idea to make that easier..

There's nothing I can't overcomplicate though Jon, its how I roll ;)

I'm very curious about the interaction between the parameters though so with things like sharpening, for example, there may be an interaction with the NR that changes things that we haven't explored. Who knows really the order in which NR and Sharpen controls come in the processing chain for example so we there may be some combination that makes the NR really viable rather than the automatically "right, lets switch that off completely for a start" setting that its perceived as. Or maybe there isn't ! And we'll probably have to throw ISO into this mix as well as that will bring another variable to the party. +3 NR at ISO640 on Cinelike D might be a whole load of different than +2 at ISO6400 on Natural and maybe not in the way that we'd anticipate it being as who knows whats going on under the hood really between the parameter value we see on screen and what is actually being affected.

 

1 hour ago, jonpais said:

Here are three more shots with the LED set at tungsten 3200 K, ISO 200, f/3.2. (1) Cinelike D, (2) Standard (3) Natural. Settings: Sharpening -3, NR -5. Even with tungsten, there's a peak in the blue channel in the shadows using Cinelike D, and with the other two profiles, there's a bump in the red channel in the highlights. Histograms (1) Cinelike D (2) Natural (3) Standard

 

And the spike is actually clipping according to Aperture.

Whats interesting is when I put the mouse over the blue square of the colour checker and you can see the RGB values (ignore the L value) under the histogram as in these screenshots.

You can see the relative difference in terms of the blue value between the profiles and yet when you do the same over the grey area of the checker they are all in the same relative balance.

Order is Standard, Natural, Cinelike D

 

Screen Shot 2017-06-16 at 11.01.16.jpg

Screen Shot 2017-06-16 at 11.00.34.jpg

Screen Shot 2017-06-16 at 11.00.54.jpg

Screen Shot 2017-06-16 at 11.06.05.jpg

Screen Shot 2017-06-16 at 11.06.31.jpg

Screen Shot 2017-06-16 at 11.06.50.jpg

Link to comment
Share on other sites

And everyone says it's the red channel you've got to watch out for with the Panasonics. Interesting. So, maybe I should be lowering saturation in camera? Also, I've read that reducing sharpening will help reduce noise, there's most certainly a correlation between the two as you say. But I haven't tested that for myself either. And in addition to the millions of possible combinations in camera, there's also the question of whether it's better to deal with noise in post with a plug in like Neat Video, or adding sharpening in post rather than in camera. So many variables!

Edit: What's unusual, for me at least, is that the Cinelike D screen shot looks the flattest, the highlights (except for my t-shirt, which is blown out in all three) appear to be the most controlled, yet the histogram is showing that it's clipping the worst. Why is that, I wonder? BTW, I'm not standing at exactly the same distance from the light in each shot...

Link to comment
Share on other sites

  • Super Members
19 minutes ago, jonpais said:

And everyone says it's the red channel you've got to watch out for with the Panasonics. Interesting. So, maybe I should be lowering saturation in camera? Also, I've read that reducing sharpening will help reduce noise, there's most certainly a correlation between the two as you say. But I haven't tested that for myself either. And in addition to the millions of possible combinations in camera, there's also the question of whether it's better to deal with noise in post with a plug in like Neat Video, or adding sharpening in post rather than in camera. So many variables!

This is how some future archeologist will find me, with the loop of incremental parameter changes still merrily being fired at a Panasonic G7 in the background.

 

skeleton-computer1.jpg

34 minutes ago, jonpais said:

Edit: What's unusual, for me at least, is that the Cinelike D screen shot looks the flattest, the highlights (except for my t-shirt, which is blown out in all three) appear to be the most controlled, yet the histogram is showing that it's clipping the worst. Why is that, I wonder? BTW, I'm not standing at exactly the same distance from the light in each shot...

One of those where we don't like what we see on a histogram but we do like what we see in the actual image.

There's a moral in there somewhere...

Link to comment
Share on other sites

  • Super Members
13 hours ago, jonpais said:

I like where you're going with this! i don't think it's as complicated as all that though. For picture profile, I'd choose the one with the greatest dynamic range, for sharpening, as sharp as possible in camera without ringing or haloing, and so on. But I still think everyone should do their own testing to see what works best for them, and if you've got an idea to make that easier...

It lives !

 

R0010944.jpg

Link to comment
Share on other sites

  • Super Members

So I've been playing with my new prodding device and....

Its interesting.

What I did was record a clip of a colour chart from lets call it a well known brand of camera that is named after a primary colour and then put the same lens that was on that camera on to a GX85 and shot it again.

Both clips were recorded onto a Ninja Inferno.

What I was doing though was monitoring the recorded clip of the posh camera on the Ninja Inferno with the scope overlay on and then switching to the input to look at the GX85 and using the prodder to manipulate the parameters live and try and replicate the signature of the recording of the posh camera. I then tweaked a few things in FCPX to match a  bit closer with the purpose of then taking those changes back to put them in the profile of the GX85 to match it better straight out of the camera.

One of these images is the film log setting from the posh camera and the other one is the GX85.

There are definitely challenges regarding noise with doing this on the cheaper camera so its obvious which is which but I've got some ideas about that.

Tweaking the values and getting instant feedback was very productive and I think this may be the start of an interesting journey.

GXcomparison.jpeg

Link to comment
Share on other sites

@BTM_Pix that is a pretty cool test. So just to be clear, you found footage online from posh camera, or you have said posh camera?

If it's the former, couldn't possible parameters in the record/upload/process/delivery chain cloud your test? If it's the latter, why are you mucking about with a GX85?

I assume the bottom grab is from the GX85? Even still it's pretty darn close. Can you turn my iPhone into an Alexa Mini?

Link to comment
Share on other sites

  • Super Members
1 minute ago, mercer said:

@BTM_Pix that is a pretty cool test. So just to be clear, you found footage online from posh camera, or you have said posh camera?

If it's the former, wouldn't possible parameters cloud your test? If it's the latter, why are you mucking about with a GX85?

Yes, I do have said camera (a RED Epic MX to be precise) so it was a completely flat test of the Film profile.

Long story about why I've got one of these but as soon as it became apparent it was useless for stills extraction they were pushing it for I did a sub-lease deal for it with a mate of mine and I've only just recently been reunited with it. I shot one event with it (mind you it was the Tour de France so it counts as 27 events I suppose) but after three or four stages it ended up being used the most expensive wide angle remote stills camera in the world as it was just way too cumbersome to use. OK for fashion I suppose but useless for sports.

 

Link to comment
Share on other sites

  • Super Members
45 minutes ago, mercer said:

I assume the bottom grab is from the GX85? Even still it's pretty darn close. Can you turn my iPhone into an Alexa Mini?

It is indeed.

Its a bit unfair on the GX85 to be honest because, as I suspected might happen with this prodder, there are some presumptions we make over its settings that might not be doing it any favours if all we're doing is correcting them with the exact inverse settings in post. And from the tweaks I was making in FCPX, I think thats exactly the case here as I was putting things in that I'd taken out in the profile so I think it can be done better.

Which obviously I'll be trying to do.

Iphone to Alexa is easy. Just wait about five years !

Obviously once you have it with roughly the same signature as the RED Film it will follow this through when you add FilmConvert etc and pick the RED Film preset. 

 

EDIT - 

 

GXcomparison3.jpeg

Link to comment
Share on other sites

1 hour ago, BTM_Pix said:

Yes, I do have said camera (a RED Epic MX to be precise) so it was a completely flat test of the Film profile.

Long story about why I've got one of these but as soon as it became apparent it was useless for stills extraction they were pushing it for I did a sub-lease deal for it with a mate of mine and I've only just recently been reunited with it. I shot one event with it (mind you it was the Tour de France so it counts as 27 events I suppose) but after three or four stages it ended up being used the most expensive wide angle remote stills camera in the world as it was just way too cumbersome to use. OK for fashion I suppose but useless for sports.

 

A couple Arri Classics are selling on eBay for $10 grand... they may make a more suitable stills camera for this year's Tour or French Open.

Link to comment
Share on other sites

  • Super Members
4 minutes ago, mercer said:

A couple Arri Classics are selling on eBay for $10 grand... they may make a more suitable stills camera for this year's Tour or French Open.

There are a couple of Fisher Price ones on there for 99 cents that would be more suitable than the RED was.

Link to comment
Share on other sites

Btw speaking of fashion @BTM_Pix I was working in NYC while the NY Fashion Show was running and I was talking to a photographer at my hotel who was there covering the show.

So of course I asked him "Canon or Nikon?" Being a video hobbyist and unfamiliar with the professional photography world, I was initially surprised by his reply... "Who cares about the camera, they're all good enough and cheap enough, it's the lens investments that dictate your body choice."

5 minutes ago, BTM_Pix said:

There are a couple of Fisher Price ones on there for 99 cents that would be more suitable than the RED was.

Yeah Pixel Vision would definitely capture the last leg in beautiful analog cassette tape. 

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

 Share

  • EOSHD Pro Color 5 for All Sony cameras
    EOSHD C-LOG and Film Profiles for All Canon DSLRs
    EOSHD Dynamic Range Enhancer for H.264/H.265
×
×
  • Create New...