GH4 Extreme noise even at ISO 200. Anyone else? (video inside)

[Razz16mm]

I've been really trying hard not to reply but here is my posture on the situation;

1.) why so many flaming posts? lots of people questioning my credentials as an experienced shooter. trust; i am. if my camera was defective, why flame me?
2.) whether my initial shot is underexposed or not is irrelevant. there will always be parts of your shot underexposed, and in those shadows, there will be noise. in some cases, even at 200 ISO it was the clients jeans that had severe noise issues.
3.) no. noise is not just "part of the picture". it's not. if my dark areas are getting extreme noise like a moth to light...unacceptable.
5.) mask the noise by adding grain? um, maybe if i'm going for that look...but nah. I won't compensate. Neat video is also not an option. It will make the image too soft.
6.) this https://www.youtube.com/watch?v=qdMypfYrKgw
7.) this http://philipbloom.net/forum/threads/gh4-noise-issue.47706/
8.) oh, and this: http://www.eoshd.com/comments/topic/6814-gh4-noise-test/

Listen guys. You are really fan girling over this camera. Be more realistic. Some people are clearly having issues. Never question the operator. I told you guys I am an experienced shooter. I simply asked you if it was defective. Most of the posts were poor attempts at flaming me. I am the creative director for a large news company. Trust me, I have shot in every situation you can imagine. I've worked with cameras with far worse noise than this one. I can understand brand loyalty, but don't let it cause you to be THIS defensive.

1600 ISO at 2.8 on a 28mm lens 120 shutter is acceptable. The sky was still dark blue. It was not completely dark out. The street lights were on. This camera performed horribly.

Sorry I posted and asked for advice. I jumped right into the fire.

1600 ISO at f/2.8 120/sec is only acceptable if there is enough light for your exposure. Did you just visually judge the light level and make a guess? Did you measure it with a meter, like the one built into the camera? Did you check it on a histogram or waveform? Based on my own experience with street scenes as you describe, your average illumination was probably 5-10 foot candles and you were two stops under exposed at least.
This is not a flame, nor am I a fan of the camera. We are just trying to communicate some essential fundamentals of photography. Like knowing what your exposure should be because you actually measured for it and know from profiling the camera how it will actually respond to your scene conditions for any given exposure setting before you shoot.
Monitors, especially on camera monitors, are useless for this.
Try some correctly exposed tests under known measured light conditions. If you are still seeing abnormal noise compared to what you see from other people's work, then maybe your camera has a problem .

 

[dustylense]

Ummm, this is still being debated? Look, that original shot by the OP is just grossly under-exposed. Second, people are trying to compare an M4/3 sensor to FF and super35. That is why I don't give 2 craps about comparing the Gh4 and the A7s in low light. The pixel density and size of the individual pixels are drastically different. I own both, plus the C100. Different cameras for different reasons.

Why does everyone want to shoot in the dark and then go on to complain about it when they see noise? I got and idea. Go out and get yourself a "ROLL" of 3200 speed film and get it developed. Tell me how smooth the image is afterwards. Hell, do it with ISO 400 film 35mm. And these days there are complaints about how noisy 1600ISO images are on digital cameras? Why? Because you're not lighting a scene.


But somehow people want a $1600 M4/3 camera to shoot in the dark. "But even at Iso200 it's noisy". Yeah, in the dark you starved the sensor of light. Try taking an HVX200 and shooting at twilight at 6db. Noisy. Yet we still complain about noise and DR. Seriously. Many people don't get how good they have it with these modern cameras. I'll call it the "entitlement generation". Or as the OP says, "Girly fanboys".

There is a reason we carry loads of lighting for professional work. I can't even fit all my various lights in the 3 vehicles plus 14ft trailer I own.

Now, finally. I've said this over and over and over here. Stop pressing your nose against the screen while reviewing your footage. Sit back. Back the F up of your monitor. You will be surprised at how noise just seems to not distract as much. But you still need to expose the images correctly.

Get out and shoot. Stop banging your head against the monitor worried about noise. Watch your exposure aids. If your waveform, histogram, false color indicators show little to no exposure, the shot will result in noise no matter what ISO you're at.

 

[dustylense]

I just read the specs in the OP's footage on Vimeo.

You used Andrew Reids settings at dusk. That is your problem. Using Idynamic, contrast all the way down, curves set +5/-5, master ped at +15? Thats why your shots are broken.

Guys! No need to shoot the Gh4 in Cinelike profiles with your contrast all the way down like its a canon 5dmk2. Stop that crap! -2 contrast at most in Cine-d.

 

[dustylense]

The sun has set in all of these Gh4 shots. Iso's 800-1600.

 

[FrameFarmMedia]

I just read the specs in the OP's footage on Vimeo.

You used Andrew Reids settings at dusk. That is your problem. Using Idynamic, contrast all the way down, curves set +5/-5, master ped at +15? Thats why your shots are broken.

Guys! No need to shoot the Gh4 in Cinelike profiles with your contrast all the way down like its a canon 5dmk2. Stop that crap! -2 contrast at most in Cine-d.

I swear people screwing with the settings too much is resulting in half the complaints you read on the internet. They included Cinelike D so people wouldn't be dialing down the standard profiles ala GH3.

 

[dustylense]

I swear people screwing with the settings too much is resulting in half the complaints you read on the internet. They included Cinelike D so people wouldn't be dialing down the standard profiles ala GH3.

Exactly. At most I'm -2 contrast -2 saturation and I still find I'm adding lots of curve in my post. If I want that milky look, raise the PED a bit. Trying to make believe you are getting more DR by flattening out your image is not gonna work. And that is why people are complaining about pasty skin tones. Because they go on this over flat profile and color setting. Try grading in camera as much as you can and one will see the benefits in their footage.

 

[Jaime Valles]

Using Idynamic, contrast all the way down, curves set +5/-5, master ped at +15? Thats why your shots are broken.

Holy cow! No wonder... If you crank up the master pedestal, you crank up the noise. Leave the curves and master ped alone and noise will be fine (given proper exposure).

 

[astromann]

Actually a few of us were talking on this thread about what our own experiences were, not necessarily about the OP anymore...just like in a normal conversation where the topic drifts a little.
Then someone comes in and (understandably) doesn't read the previous pages and then says something about the OP problems "exposure" etc.., when that has already been mentioned several times on previous pages...
and many of us were not exactly talking about that anymore, we were talking about our own experiences (a bit off topic I know).
Then the OP comes on and is (understandably) pissed off because he feels that on page after page people are criticizing him and his work, but thats not necessarily the case, its just that we want to know what our own issues
with noise and ISO are. But thats forums for you, no one really reads the previous pages and the whole thread goes here and then lurches back to there.

Personally (aside of my own experiences with the settings and the camera) what I was interested in......is there a difference in the cameras at all?

I used to import top of the Line Fender Stratocaster guitars from Japan (often comparable to Fender USA custom shop models)...and they are a lot simpler (electronics wise) than a camera like the GH4.
I A/B'd several models and I can tell you, they sounded different, some were more full and rich sounding, others were not, now the PUPS and electronics were done with the exact same factory tolerances and windings,
yet they sounded different (and these were built in Japan under very strict quality control) I believe the GH4 is built in China, its harder to get consistent quality control there.

Now the OP posted this link amongst others http://philipbloom.net/forum/threads/gh4-noise-issue.47706/

Here the original poster Patrick Fahy said this in his first post
--------------------------------------------------------------------------------------------------------------------------
The noise is a little nuts on the window sill. The ISO here is at 200, and I'm in the CinD profile.
I tried again with standard profile and got the same issue. Tried my GH3 with the same lens in the same place and got no noise.
Is anyone else having this issue?
--------------------------------------------------------------------------------------------------------------------------
But further down he posted this (for anyone who has not read that link properly)
--------------------------------------------------------------------------------------------------------------------------
New GH4 came in with a much better picture.
It was a flawed sensor and B&H accepted my return.
My tests were under exposed and in standard profile with everything matched up to my GH3 on purpose so I could compare with my GH3,
which had none of the noise issues at identical settings. It was just an out-of-the-box issue.
Haven't put the camera down since I got it, so don't let this test sway you. GH4 rocks.
Thanks for the help,
--------------------------------------------------------------------------------------------------------------------------

I find this interesting, we know that the OP vid was underexposed, and also Patrick Fahy was underexposed, but it appears that there was a another problem, noise at ISO 200
could definitely be attributed to something else,and from this post it definitely appears to be the case.
So apart from Andrew Reids settings, this could be the reason that some are having no real issues and others are having issues even at low ISO's

BTW @dustylense
Some really nice caps there, nice night shots and really great lighting!!

 

[dustylense]

I have a Japanese and a USA strat. I prefer the Japanese version. The Mexican plant versions are getting quite good too. But keep in mind, there are no ISO standards to guitars. The very same coiled pickup will sound quite different on another neck or wood make. No two guitars play the same.

 

[astromann]

I have a Japanese and a USA strat. I prefer the Japanese version.
The Mexican plant versions are getting quite good too. But keep in mind, there are no ISO standards to guitars.
The very same coiled pickup will sound quite different on another neck or wood make. No two guitars play the same.

Nice to hear from you dustylens
Ha ha Now we are really going off-topic.
Yep the Japs build some awesome guitars for sure, dating back to the Gakki factory...I tell you one of my favorites (outside of Fender Japan) is Edwards (ESP), they make a frikkin awesome Les Paul.
I used to own the original (believe it or not...back in the day) a 59 Les Paul Cherry Top (worth $.....???) these days!!
And the Edwards Les Paul is very close to being the same...and a lot lighter.

Here is a vid of me using it here DI'd but you can hear the sound...the general idea
https://www.youtube.com/watch?v=K1gj_PCTWhw&list=PL7CZoAVTSdewZEerIOmswxPHVscs0jJen

BTW Yep Those Mex Strats used to be s*%te! but these days they are really getting loads better, with some great models..Jeez I am definitely tempted!!

 

[j1clark@ucsd.edu]

I find this interesting, we know that the OP vid was underexposed, and also Patrick Fahy was underexposed, but it appears that there was a another problem, noise at ISO 200
could definitely be attributed to something else,and from this post it definitely appears to be the case.
So apart from Andrew Reids settings, this could be the reason that some are having no real issues and others are having issues even at low ISO's

BTW @dustylense
Some really nice caps there, nice night shots and really great lighting!!

The point of my posts, is that if one finds some 'unexpected' result... rather than post a 'happy shot', set up a test, and have some examples with standard charts/targets, to show the 'noise', or 'the lack of resolution', or whever is the unexpected feature... early on people were complaining about 'jello' when they 'whip panned' their camera... well... even for Film film there were standard movement speeds to avoid 'artifacts'. And these complaints may be legitimate, or due to the cost of the camera, just part of the device's capabiliites... or lack thereof...

For example a lot of talk about 'moire' effects... well... guess what... that's expected for cameras like DSLRs that have an optical filter designed for 4K (or so...) still images, but when used in a 1920x1080 mode for video is not correctly tuned to roll off 'high spatial frequencies' that contribute to the 'moire' effect. And just as obvious, for a camera that is capable of 4K moving pictures, there should be less of such effects, because the video spatial parameters are close to the still parameters.

The solution is to use a 'low pass' filter in post... but that 'degrades the image'... well duh... but an 'optical low pass' filter would be tuned to do approximately the same thing... now there may be a small issue of exactly when the filter is applied, and it may be 'better' to apply it before digitization... but then one still does not get the absolute 'sharpness' that the still image is capable of... One solution of course is to use a 'soft' lens... but anyway...

The basic message is test like hell... and then test some more...

For ISO, get step wedges, or even a grey card standard, and then demonstrate the 'noise'. Read up on how ISO is determined... etc... for digital systems.
For resolution/moire... get a pattern target and demonstrate the aliasing... etc...

Some people may not be willing to go through that sort of hassle, and it is a hassle... but I think that is in part what it takes to become a knowledgeable camera person.

 

[Ian-T]

<<sigh>>...my Strat broke and is just collecting dust...

 

[astromann]

The point of my posts, is that if one finds some 'unexpected' result... rather than post a 'happy shot', set up a test,
The basic message is test like hell... and then test some more...
n.

I found your posts and dustylenses posts very informative and helpful, so thanks for posting that information.
And yes I agree 100% with this, testing can be a bit of a pain in the arse, but you get to know the camera and its limitations that way for sure.
I cant get out right now to shoot the footage I want...so I can hopefully put the time to good use by doing a lot of tests.
Cheers

 

[VMT]

You are really fan girling over this camera.

I often wonder why there are no "girls" on this sausage forum. Now I know.

 

[Fahnon]

I've been really trying hard not to reply but here is my posture on the situation;

1.) why so many flaming posts? lots of people questioning my credentials as an experienced shooter. trust; i am. if my camera was defective, why flame me?
2.) whether my initial shot is underexposed or not is irrelevant. there will always be parts of your shot underexposed, and in those shadows, there will be noise. in some cases, even at 200 ISO it was the clients jeans that had severe noise issues.
3.) no. noise is not just "part of the picture". it's not. if my dark areas are getting extreme noise like a moth to light...unacceptable.
5.) mask the noise by adding grain? um, maybe if i'm going for that look...but nah. I won't compensate. Neat video is also not an option. It will make the image too soft.
6.) this https://www.youtube.com/watch?v=qdMypfYrKgw
7.) this http://philipbloom.net/forum/threads/gh4-noise-issue.47706/
8.) oh, and this: http://www.eoshd.com/comments/topic/6814-gh4-noise-test/

Listen guys. You are really fan girling over this camera. Be more realistic. Some people are clearly having issues. Never question the operator. I told you guys I am an experienced shooter. I simply asked you if it was defective. Most of the posts were poor attempts at flaming me. I am the creative director for a large news company. Trust me, I have shot in every situation you can imagine. I've worked with cameras with far worse noise than this one. I can understand brand loyalty, but don't let it cause you to be THIS defensive.

1600 ISO at 2.8 on a 28mm lens 120 shutter is acceptable. The sky was still dark blue. It was not completely dark out. The street lights were on. This camera performed horribly.

Sorry I posted and asked for advice. I jumped right into the fire.

Have you seen this post? It has a whole bunch of noise tests with video and text commentary and includes a lens cap test. Might help.
http://bkmeditor.wordpress.com/2014/07/16/panasonic-gh4-iso-noise-tests/

Some relevant info from that page:

"...settings were Cine D [-5,-5,-5,-5,-5,+2] Shadows/Highlights 0/0 and Master Pedestal +15...settings were a disaster with this test. What I found overall was that boosting the Master Pedestal increased noise across all ISO settings...I found noise even at 200 ISO."
"In contrast using Cine D [-5,-5,-5,-5,0] Shadows/Highlights 0/0, and Master Pedestal 0, I found much less of an issue with noise."

 

[Dan Keeble]

you can see the how significantly sharper the GH4 4K footage looks when downscaled to 1080 vs the close-up shot with the GH3. Granted, the GH3 was using an old SD broadcast B4 zoom lens and the GH4 was using the Panny 14-140 f/4-5.6 lens, but still, the difference is stunning.

loving this. GH4 has got to be my next camera after seeing this.

 

[RickyNorris]

Can any of you guys say pretty definitively what iDynamic and Iresolution actually does to the image? I've done some testing but not fully decided yet what to make of it. Also really curious if some of you have your NR at -5 and have clean images at 200? I've just found that 0 cleans it up so much better without much softness sacrifice, with everything else dialed at -5, and curve unmodified.

 

[themac]

iDynamic is a one button solution to increase your dynamic range by not blowing out the highlights and raising some details in the shadows, in some cases I think it can be helpful, how it does this I'm not sure. iResolution is one that is probably unnecessary, shooting in 4K down scaled to 1080p buys you so much detail and resolution already. NR didn't seem to make a big difference for me so I just left it at 0 and hope that Panasonic knows best for this.

I've found that shooting on CineD with -3 saturation works 90% of the time. Sometimes I leave saturation at 0 if I intend to have popping colors. Notice I don't drop the contrast at all as I think it's plenty flat enough for a compressed 4:2:0 8bit codec, in my opinion more unnecessary noise was always added at the cost of some detail in the shadows that I'd just crush again in post. The tweaks to pedestal, shadow/highlights and iDynamic are very situational but can help, I'd definitely advise against using those extreme tweaked settings for majority of your shooting as it'll probably do more harm than good.

 

[astromann]

@RickyNorris
I have recently done a load of tests.
Basically it pans out to this...(Please note...If anyone reads this, I am not really refering to the OP, we are in a general discussion about noise and this is a response to RickyNorris)

First at ISO 200 NR-5 or Zero...cant see any noise in either, even in dimmer light (where the histogram is to the left mainly)
Ditto for ISO 400 (same tests...No Noise) even if NR is at -5 or Zero
I even see any at ISO 800, but it may be possible to get some happening if I really make the conditions dark and gloomy, its possible, there probably has to be different tests, maybe a dark room with something white in it.
Then at ISO 1600 some fine grain noise (monochrome) is visible if the sensor is really starved for light, otherwise there is very little (miles less than the GH2 for example...even with the best hack)
At ISO 3200 Noise starts to appear (still fine grain) but there is none visible if there is sufficient light, again the sensor has to be starved for light...(Histogram hard left..exposure well into the minuses) for noise to appear.

At 64 yes there is noise, but in my camera its totally acceptable, its still fine grain, but if you lift the gamma up in the shadows, then blotches start to appear, but this is in very dark conditions, where to your eyes its pretty much black without details.
Other than that I could use 6400 ISO definitely, I simply am not seeing 800 ISO as being the highest this cam can go without the noise getting in the way, this was mentioned by James Miller, and possibly Philip Bloom (from memory)
my experience is different than that...and I have been a graphic artist, 3d designer for a long time...so I am fussy as I have to create my own scenes from scratch...with lighting, sub surface scattering in the skin etc...etc...
so I have had a fair amount of experience in this.

Most of the tests were done at 2.8 ...I have faster lenses, but I used the 12-35 at 2.8 for all the above.
As far as Noise reduction at -5 vs Zero and winding the sharpening down, to me the differences were subtle, not that noticeable (on a 27' 2560 x 1440 screen) viewing at 100%.
I even shot at night and boosted the shadows to +5...still no real noise, in fact I liked the result...other than my WB was off...so it looked a little golden or yellowish tinge.

I totally realize that others have had a vastly different experience, I have no criticism at anyone, or even a debate, I think at the end of the day, you have to work out the camera for yourself.
But its good to compare results at times, it often helps.
I will say that...if the noise was blotchy, multi colored, crawling etc...I would be climbing the walls and complaining loudly, but I am not getting that.

There are some example caps below in dark conditions,(All at 2.8 Aperture) I only lowered the ISO to 400 at the lowest, Histogram was really hard left, and they looked much better at 4K,
the light was actually darker than the ISO 400 Images I put here,only when a car went by was there some good light, the street lights were not very good at all.

Plus I pushed the crap out of the shadows in the 1600 and 6400, I think the NR was at Zero or -2 maybe....IR etc...was off and settings were Natural.
WB is off I know...I think it was 3300 from memory, therefore the yellow cast LOL!!

So far my only gripe with this camera is that at 4K sometimes movement looks just a little strobey...but its not that bad really especially with OIS, its definitely not anywhere near A7s territory.
Hope this helps

 

[RickyNorris]

Thanks @astromann, so comprehensive. Glad to see you finally got yours and have had some time for fun. I've loved mine so far as well, been absolutely great for fashion and lifestyle content. Funny enough today I was cutting a project I shot earlier this year on my 5D3 just before the GH4 came out, and noticing some noise I never had before. Perspective is amazing. And all the footage looked so soft in comparison to what we get to have these days, and mind you I loved that camera.

Another thought came to mind, a few years ago when I was just getting into shooting docs, I shot a bunch on a Panasonic HPX170, which was an awesome camera at the time. I didnt try and super flatten the image but just got it looking good in camera and so much footage from those days I still love. Taking a similar approach with the GH4, I want the image I bake in there to be great. And so far its wonderful.