Hmmn? Cinema or Standard? Don't understand...
GH2 Mega GH2 ISO Noise test w/ Quantum beta V9b. Interesting results.
- Thread starter Tim Joy
- Start date
Maybe he means 24p Cinema mode with Standard profile
Tim Joy
Veteran
Some other findings while I compile another massive test that involves Highs, mids, lows, gradients, over exposures, different iso's, profiles and hacks all in one! ughh.
EDIT- Of course pics are resized by your friendly dvxuser-bot, but still gets the point across.
EDIT- Of course pics are resized by your friendly dvxuser-bot, but still gets the point across.
Attachments
Now I'm just confused. What is that test supposed to demonstrate?
Tim Joy
Veteran
Good question. It's changing a couple variables, but it looks like the nasty ISO 320 is good in Standard profile, but not in Smooth. (At least on firmware 1.1)
BUT, turning Noise reduction up to +2 actually INcreases noise in the shadows.
In the iso 640, HBR looks just like NR +2 on 24p, and iContrast and iResolution increase the noise.
3200 is just noisy no matter what
But I don't like the 'sharp edgyness' of the noise pattern with HBR mode on.
BUT, turning Noise reduction up to +2 actually INcreases noise in the shadows.
In the iso 640, HBR looks just like NR +2 on 24p, and iContrast and iResolution increase the noise.
3200 is just noisy no matter what
But I don't like the 'sharp edgyness' of the noise pattern with HBR mode on.
Last edited:
Cosimo Bullo
Veteran
Sorry, I meant to say look at cinema at 1600. It's an amazingly clean profile.
That said, I think Standard, as 'boring' as it sounds, is the go-to profile for the GH2. And I'm now close to sure that the camera is in fact native 320 ISO.
I agree with a previous poster that 400 and 640 have virtually identical noise patters at the same exposure.
That said, I think Standard, as 'boring' as it sounds, is the go-to profile for the GH2. And I'm now close to sure that the camera is in fact native 320 ISO.
I agree with a previous poster that 400 and 640 have virtually identical noise patters at the same exposure.
Cinema is "clean" because there's a lot of NR going on in the actual profile itself. I think standard is the most linear profile you can get. And, if you do comparisons at the same ISO's, you'll notice that some are either way too sharpened or have NR happening that's unpleasant.
I like the way Cinema looks, but I grade anyway, and everything thhat Smooth and Cinema can do I can accomplish in post, anyway. CLeanest image to start, then options later.
I like the way Cinema looks, but I grade anyway, and everything thhat Smooth and Cinema can do I can accomplish in post, anyway. CLeanest image to start, then options later.
Uwe Lansing
Veteran
I couldn´t resist to take my GH2 and a small tripod with me this morning (jogging/walking). Although it was pretty rainy and stormy I headed towards the Elbe and took some comparison shots between iso 320 + 640 in real life. As for me it´s fairly clear which one looks better to my eyes (of course, at the original files it is more obvious). But please form your own opinion:
http://vimeo.com/35455135
Password: noise_test
http://vimeo.com/35455135
Password: noise_test
Fresno Bob
Well-known member
Nice test, I am pretty sold at standard profile at 640 iso as my default native iso. It is pretty awesome though that 1600 is pretty clean, that's a very very light light kit to work with. It's weird to be able to get by with a few LED's. Actually makes controlling the shadows and creating a mood a little more challenging.
How were you controlling the exposure between shooting 320 and 640? I'm probably wrong, but the right way to do it would be with shutter increments and not NDs or Stopping down.
Light hitting the sensor needs to still be the same amount to properly test. If 640 is cleaner in circumstances where the light coming in is consistent, it wouldn't be in comparison to 320 if you chopped the amount of light out before the sensor.
Uwe Lansing
Veteran
The first scene was shot with a different shutter speed (1/25 + 1/50). The other two with different apertures. I suppose that it doesn´t make a huge difference whether you regulate the amount of light with the shutter or the aperture - the amount of light hitting the sensor will be the same. But I´m not sure either...
It makes a massive difference. MFT(GH2) does not use the mechanical shutter for video, it's electronic. The only way to regulate light before the image is aperture or ND, the shutter's software controlled simulation.
By adjusting aperture you're adjusting how much light's hitting the sensor, and it will definitely skew the results. The only real way to test (as far as I can figure) is to use the shutter to balance the exposure to the meter reading and then further tweak in post to exacts. That will give you your answer on which one is cleanest.
And don't forget the ISO bug isn't just limited to going up and then down again: when you turn the camera off and on once more you have to do it as well, no matter which ISO you're on.
Uwe Lansing
Veteran
Sure, I did it that way.
So, the first scene gives a right result. But, as I already said, I suppose the results of the other 2 scenes wouldn´t differ that much regarding the IQ and the amount of noise if I regulated it with the shutter speed. But perhaps you can show us the difference in some comparison shots?
I'm not saying which one is cleaner, by the way, just putting the info out there on how to test effectively. The reality of it? I'm not gonna try to cap light with a higher shutter, I'm gonna do it with ISO first then ND from now on. Which, working with RED, I went the opposite way because I'm not dealing with 8-bit 4:2:0. In this case, I'm gonna pull as much light in as possible then balance in post.
In summary, I'd rather shoot 400 ISO with a Polarizer outdoors than shoot 640 with a Pola and ND.
Cosimo Bullo
Veteran
I know, it's a bit ironic, but I think the native ISO on the GH2 may be 320, dangers and all. The 320 bug, IMHO, is best worked around by using a custom menu that always starts the camera at 400ISO (another good setting, but with, it seems, identical noise to 640ISO) and click to 320 at start. We're cheating with out GH2s by putting paper tape with 320(!) conspicuously reminding us to make the change at start up. 640 looks great too, not saying it doesn't. 320 seems best though. Breaks up the banding a bit and keeps noise very low.
Tim Joy
Veteran
Well, after doing some pretty extensive tests with much more real-world scenarios I found that the amount of noise that would show up was highly sensitive to exposure and the correction filter I would place on it to enhance the noise. I could almost make 640 look like 320 and vice versa with just subtle changes in the enhancing filter.
Since it is near impossible to match exposure any closer than I did, and the results kinda told me that there's no major difference between noise patterns of the various iso's, other than the fact that they get noisier as you go up, I started to think there is not really a "best" iso.
I think I will just use what I need to.
The biggest two things I took away from all this is 1) DON'T underexpose this camera in low light, and 2) the hacks are awesome because they show the grain and I like it so much more than macro blocks.
Since it is near impossible to match exposure any closer than I did, and the results kinda told me that there's no major difference between noise patterns of the various iso's, other than the fact that they get noisier as you go up, I started to think there is not really a "best" iso.
I think I will just use what I need to.
The biggest two things I took away from all this is 1) DON'T underexpose this camera in low light, and 2) the hacks are awesome because they show the grain and I like it so much more than macro blocks.