GH2 vs Sony NEX 5N

brianluce

@EOSHD
I don't have a good monitor but parts of the book on the 5n jpeg look clipped. At the least, I can make out detail on the book on the GH2 that are invisible on the 5n. So I don't think the exposures were the same.

Ian_T

@brianluce Exactly. I bet if he dialed down the exposure to match then we would also see issues in the dark areas...as we seen in the GH-2. I appreciate the tests...but something's a little off.

Also, a 1.62 vs 1.85 crop factor visually is not that big a difference (at least it didn't stand out with my GH-1 vs my 7D...so it should be the same here with the Sony cams). Remember..the GH-Xs are not the typical sized M 4/3 sensors. They are actually bigger.

EOSHD

The phrase isn't in the review Brain! I should know I wrote it! Do a ctrl-F. I mention 'Gradated tones' to describe gradation between two tones. I don't see a problem. This kind of pedantry I can do without.

You don't judge exposure from highlights. Take the mid tones as your reference for exposure. You will see the GH2 shot is actually the one that is slightly more exposed than the 5N yet you have the problems in the shadows... go figure. There's more detail on the book because the GH2 resolves more detail, simple. Highlights are clipped at the same point on both.

I am using both GH2 and 5N in my projects they both have their own unique strengths and weaknesses.

Ian_T

Ok so here's what I mean by something being a "little off." When I look at both of these images the first thing that sits out is the FOV. They are almost identical (as a matter of fact the GH-2 is a little wider here). Now we know that the GH2's M4/3 sensor is slightly smaller than the Sony's APS-C which then means it should yield a "slighly" smaller FOV right? But because the FOV in the two shots are "almost"identical (Oxymoron anyone?) then it tells me that the shooter had to compensate for that by repositioning the camera and adjusting the iris. So already this is not an "equal" test of the two cameras.

@EOSHD you said " I exposed to just about protect the very bright highlights on the book in the sun, base ISO"

What does that mean by the way? Did you give them both the same shutter/aperture etc.? Or did you just adjust each of them to the best possible exposure for the book to protect its highlights (as you said)? I ask because you can see the aperture is opened up more on the Sony (blurred background slight overexposure etc.). People will immediately conclude that the bigger sensor is the reason for this but IMO that would be wrong.

Someone correct me if I'm wrong (and I'm sure someone will) but if given the exact same aperture/shutter speed etc. then the background blur and overall image should look indentical...except for FOV (where there would be a slight crop on the M 4/3's.

With all that said...this is why I believe we see a slightly overexposesd image on the Sony where I would expect cleaner shadow details vs the struggling shadow areas of the GH-2 and background (trees) more in focus of the GH-2 (including the better exposed book etc.).

I could be wrong though. :-)

By the way...was this a hacked GH-2 or was it stock?

stefanos

@EOSHD
thanks for 2nd test. I suspect the main reason you're seeing such macroblocks is because of the film mode. I have conducted my own tests of the GH2 film modes, and cinema is by far the worst when it comes to shadows (and highlights, but that's another story) . Cinema mode is very destructive with shadows. Nostalgic fares the best, smooth is also good. even standard is better than cinema. If you find the time, it would be greatly appreciated if you could repeat a similar test with the two cameras, the GH2 set to nostalgic and/or smooth (hacked with cbrandin's or driftwood's settings). Then bring the files from the cams into after effects, open color finesse 3, go to RGB and then shadows, increase the shadow "gamma" a lot, then copy the effect (Ctr+ C) and paste it on the footage of the 2nd cam. And finally export 2 png frames, 1 from each cam.
thanks

B3Guy

Is the GH2's FOV different with M4/3 or is their interesting use of sensor real-estate use (not using sensor corners) actually making them very close?

Ian_T

@B3Guy I think how it goes is a typical M 4/3 sensor measures 18 mm × 13.5 mm (22.5 mm diagonal), with an imaging area of 17.3 mm × 13.0 mm (21.6 mm diagonal)<<referenced Wiki >>

That, I believe, describes the AF-100 and M 4/3 cameras like it.

The GH-X series of cameras however have a bigger M 4/3 sensor, 19 mm x 10.7mm (21.81 diagonal) to be exact. If I'm not mistaken the GH-2's sensor is shaped in such a way where it uses the "entire" sensor for its 16:9 video. (meaning....no cropping).

EDIT: According to Wiki "To date, the multi-aspect sensor is common only to the Panasonic GH1[5] and its successor the Panasonic Lumix DMC-GH2.[" FYI

All too often people downgrade the M 4/3 sensor vs APS-C but I think they do so without having the bigger picture in mind (ok....pun intended). But the APS-C cameras have to actually "crop" the top and bottom of their sensor to get the 16:9 image. So...in the end...the GH-2 vs APS-C's crop factor is not a big rift in terms of FOV. There is a difference...but not that much (at least not like comparing a FF 5Dll to just about any other camera on the market that shoot video).

If one is looking at taking still pictures with m 4/3 vs APS-C....then yeah....there is a big difference. But in the video world...don't let people fool you. They are "almost identical" looking. This is one of the reasons why the term "crop factor" is totally useless when comparing large sensor cameras that shoot "video."

LPowell

@Ian_T - "But because the FOV in the two shots are "almost"identical, then it tells me that the shooter had to compensate for that by repositioning the camera and adjusting the iris. So already this is not an "equal" test of the two cameras."

FOV on the GH2 in video mode is about 20% narrower than APS-C cameras. In comparing different types of cameras, I think it's more useful to set them up to produce comparable results. Technical settings are not always calibrated the same on different cameras - in particular, ISO varies greatly among cameras and is no longer a reliable industry standard.

Here's a link to a comparison I did between the GH2 and the APS-C Nikon D5100:

http://www.personal-view.com./talks/discussion/841/low-light-video-test-gh13-vs-gh2-vs-nikon-d5100

What I did to normalize this test was to use the same lens set at a fixed aperture and shutter speed. To normalize the FOV, I set the lens focal length to 22mm for the GH2 and 28mm on the D5100. This allowed me to keep the lens in the same position, so the perspective remained unchanged. To normalize the exposure, I had to use different ISO settings on each camera, and to normalize the white balance, I had to make individual adjustments to the cameras' film modes.

I think this approach produces more useful results than a strictly technical comparison. In the end, I'll want to intercut footage from different cameras and this helped me determine how well each camera can produce the kind of results I'm looking for.

Ian_T

@LPowell You are absolutely correct about the unreliable ISO values.

But still...if I were new to all of this and seen those images for the first time how would I be able to tell that one camera is an M 4/3 and the other is an APS-C? Because I don't see a 20% difference in their FOV then I would have thought they were the same type of camera. Setting them up to produce comparabe results is one thing but setting them up to show the differences is another (IMO).

I only intially replied to that post because of EOSHD's comment about the noise in the GH-2's lowlight area vs the overexposure of the SONY. When you change all those other variables then something's going to have to give. I think if everything was shot exactly the same (position in room, lens, f-stop etc.) then we could make a better judgement on the potential noise/overexposure issues. We might find out the main issue is just the 20% difference in FOV (ok...and resolution).

To each his own I guess. :-)

LPowell

@Ian_T - "But still...if I were new to all of this and seen those images for the first time how would I be able to tell that one camera is an M 4/3 and the other is an APS-C?"

You can't. In evaluating test results, you always have to rely on the integrity of the tester.

rigs

here is a cool low light lighter test of the iso ranges. Love the beer

brianluce

@EOSHD
What do you mean you can't judge exposure from highlights? Of course you can. If that were true, then if something was clipped, you're not supposed to judge it or worry about it? If I'm looking to judge dr I look from the brightest to the darkest areas and see how the image holds up at respective extremes. That has nothing to do with composition of course, because sometimes you want or have to have areas under or over exposed.
Like I said, I'm just viewing those samples on a Macbook, and on my Macbook that book is more exposed on the 5n. It's nothing about resolution, where I see white on the 5n, I see print on the GH2.

EOSHD

I think that guy had been drinking it before he shot that :) The 5N is capped at ISO 3200 for video recording. He has shots in there as high as 12,800 and they are nothing of the sort. If you set ISO above that, say 6400 the camera reverts back to 3200 when you hit record.

Tests also rely on the integrity of the viewer. Be educated enough to know what you see. It's easy to tell a bad test from a good one, always draw your own conclusions. I am a filmmaker, I am not really interested in scientific testing methodology so I don't shoot charts but I know how to reveal a certain aspect of a camera's performance relative another... you don't always need the perfect identical set of parameters to show that. It is not sloppy testing to do otherwise, it is cutting to the chase and taking out the irrelevant details.

I am still interested in the macro blocking on the GH2 shot in the lows and what is causing it. My own best guess is that there isn't enough information in the lows (either subtle signals from the sensor have been truncated somewhere in the image processing pipeline before it gets to the final encoding stage or the sensor just hasn't picked up any details in the shadows) and so the encoder gives it short-shrift and a low priority, hence the macro blocking.

Bear in mind the stills attached above are 1:1 crops from the centre of an AVCHD still frame and it was a 2 minute test, don't read anything else into it other than for what it was designed: to show dynamic range in video mode of the 5N and GH2 and how clean the encoder is in the shadows.

Look @brain I tried to get the exposures as close as possible, you are talking about a TINY difference here. It doesn't account for what you see in the shadows on the GH2 if anything as I already said the GH2 is slightly (very very slightly) more exposed than the 5N so it should be picking up MORE in the shadows than the 5N not less. I feel this is a typical case of someone trying to win an argument where there shouldn't even be one, by picking up on semi-relevant details and amplifying them until they obscure the truth!

brianluce

@Ian_T
It was a hacked 44mb/C.Brandon setting.
@EOSHD
I'd appreciate it if you spelled my name right. It's not funny.

EOSHD

Typo... sorry :)

Ian_T

@EOSHD LOL..I was going to say none of those ISO changes got any brighter. You're right...maybe he just downed all of those beers.

cbrandin

Come on guys - stay polite. By the way @brianluce, you did misspell my name - and under the circumstances I think that is very funny!

Chris

Ian_T

...until your post I had no idea anyone's name was mispelled. I guess I have to read more carefully.

Anyways...thanks for the settings brianluce.

Butt

@all
sorry - what are you doing here:
is pixelpeeping-bullshit

EOSHD

Any idea what's going on in the shadows @cbrandin?

cbrandin

Cinema gamma mode is absolutely the worst film mode for shadows, by a long shot. You ran this test using the GH2's worst possible setting for dark areas. If you look at the GH2 film mode curves you'll see that cinema gamma severely crushes shadows - when they are boosted back to normal they will look terrible because of severe quantization and the corresponding macroblocking.

Chris

adventsam

Looks like 5n has gh2 beat, 50p is real cool, 25p is ultra cool in pal world. Plus the 5n is plenty sharp for movie and much more cinematic, gh2 looks video'ish and a stop darker in iso, maybe more? love my gh2 but this is a head tuner, size, iq(especially stills) are in another league.

stefanos

@EOSHD you were asking about why does the shadow lift looks so horrible, I gave you my opinion a few posts behind, and was ignored:)
anyways, in case you might be interested, here is my previous post again, hope it might help.

"@EOSHD
thanks for 2nd test. I suspect the main reason you're seeing such macroblocks is because of the film mode. I have conducted my own tests of the GH2 film modes, and cinema is by far the worst when it comes to shadows (and highlights, but that's another story) . Cinema mode is very destructive with shadows. Nostalgic fares the best, smooth is also good. even standard is better than cinema. If you find the time, it would be greatly appreciated if you could repeat a similar test with the two cameras, the GH2 set to nostalgic and/or smooth (hacked with cbrandin's or driftwood's settings). Then bring the files from the cams into after effects, open color finesse 3, go to RGB and then shadows, increase the shadow "gamma" a lot, then copy the effect (Ctr+ C) and paste it (Ctr+V)on the footage of the 2nd cam. And finally export 2 png frames, 1 from each cam.
thanks "

Edit: just saw your post @cbrandin :) hopefully Chris' confirmation will be taken more seriously

EOSHD

Thanks Chris, I was not aware Cinema Gamma crushes shadows, I ran it because Nostalgia (my usual preference) was giving me a poorer roll-off from highlights and blowing them out easily. So Nostalgia boosts shadows and highlights, whilst Cinema Gamma protects highlights but at expense of shadows.

This test is a 2 minute test. An experiment. I'll do the same with different film modes (and on the 5N also) to see if they reduce the effect. This was only a quick thing, I wasn't even going to post it but it turned out to be quite revealing.

Don't make any final conclusions guys. I haven't yet either.

But the 5N does seem to hold up to grading well.

EOSHD

Sorry @stefanos I genuinely missed your post, wasn't my intention to ignore. Had a busy few days :)

Howdy, Stranger!

Categories

Tags in Topic

Top Posters