Personal View site logo
GH2 vs EX3 stills...
  • I'm not really a pixel peeper, but here they are anyways... for whatever it's worth.

    Note: I'm in the middle of a move right now, and my new place has no internet currently, so these still are all I can upload for now. I'll do the clips later.


    EX3
    image

    GH2
    image

    EX3
    image

    GH2
    image

    In my personal interpretation, I say the GH2 resolves a bit more detail. But they are very close. Both are true 1080p cams. Whoever was saying the GH2 could only resolve 600 lines was out of his mind. 600 lines is even below a 7D.

    The EX3 in these shots looks like it has a bit more highlight control... but it was more likely caused by user error since I did this pretty quick and the EX3's monitor is much more faithful to the final image. I'm sure you can expose for better highlights, plus, there is much more shadow latitude that's not infested with noise like the EX3.

    These are also still images, so in motion they look a bit different. IMO, the GH2's image in motion looks better than it's stills, and the EX3 actually looks worse due to the huge amounts of noise. When moving... the GH2 looks like motion still images, EX3 looks like noisy video.

    That's all for now.
  • 31 Replies sorted by
  • @bwhitz
    Cheers! That's good news for us fanboys.
    However your test doesn't show wheter it will make a nice 100 foot projection or not.. ;)


  • @RRRR "However your test doesn't show wheter it will make a nice 100 foot projection or not.. ;)"

    ... that part is up to you. ;)
  • As long as its not pushed, the GH2 should look close to a Viper projected.

    Have a colorist milk the image from a good DP, print to 35mm and then scan at 4k, you'd have an incredible projected image.
  • @plasmasmp
    I'm pretty confident the gh2 projects beautifully..
    it was a bit of an internal joke due to a previous topic in which the actual line resolution of the gh2 (and wether it would project well) was brought up.. bwhitz's test, I think, came as a result of that.
  • Nice test, thanks for posting @bwhitz!
  • @bwhitz,

    That's quite the difference, really. I own an EX1 and the GH2 with driftwood's hack leaves the EX1 on the shelf.
  • I have owned GH2 and EX3. I think it's depends what kind of event you are shooting. If you have a time (for control) then GH2 is a winner. If you don't have a time (run and gun style) then EX3 (EX1) is a winner.
  • @Kolbasa

    Yea, exactly. They're both good cameras... just use the right one for the project. I think most people here are after the highest quality cinema-picture, so the GH2 wins in that category.
  • Thanks for the test, what were you settings on the gh2?
  • True, if I was tied to mainly event work, The EX1 and EX3 are work well and look great here.
  • @danyyyel

    Shot on 'cinema' with everything turned to -2. ISO 640

    Lens was a Sigma 30mm 1.4 wide open (which is also impressive) with the Kipon m4/3 to EOS adapter.

    On another note... Is cinema the best setting for DR or is smooth actually better in the end?
  • I wish my 7d was that sharp.
  • @bwhitz Nostalgic has the highest DR on the GH2. Smooth is a close second with standard and doesn't have the tint. Cinema is actually one of the worst.

    http://www.dpreview.com/reviews/panasonicDMCGH2/page12.asp
  • But Cinema is less noisy in low light than Smooth and Nostalgic. So even if dynamic range is worse in Cinema it may actually look better. But then maybe I could just as well decrease iso and use Nostalgic at a lower ISO?
  • These pictures reveal how poor cinema mode is - even the ex3 outperforms the gh2 when stuck in 'cinema' netherworld.
    There is a myth that cinema is less noisy - it is not, it just forces the use of the clean part of the dr (next to blow out) by throwing out the shadows.



    Smooth -2 always
  • @ronnylov It appears less noisy because the shadows have been crushed compared with other film modes; crush the blacks in nostalgic and it will look less noisy too. Whether it looks better or not depends entirely on the dynamic range and contrast of what you're shooting.

    @sage You don't always want a profile with a high dynamic range and very low contrast; if you're shooting a scene with a very low DR you won't be taking advantage of the bit depth if you have lots of headroom on your shadows/highlights; your profile should reflect what you're shooting.
  • I am aware. There is more to consider -
    Quantization is worst in the shadows - a much lower effective bit depth stored (even post hack)
    While smooth pins values well above black, the midtone luma still has much contrast on gh2 (see recent PB test).
    You can have two 'my films' loaded for varied contrast in smooth, ready to go - but in reality the theoretical point is *not played out, my 32 bit grader (common tool now) does just fine with the hacked smooth -2 footage. In contrast, loss of dr due to high contrast settings is a real problem I have seen too often.
  • In general, I don't care much about these types of comparisons (thanks for posting though).

    I'll just note that the assertion that "cinema" is a "netherworld" and that "smooth -2 always" are personal preferences, not facts. It is true that cinema has less dynamic range relative to nostalgic and smooth, and that it does crush shadows more. Some people assert that cinema has a ugly, yellow-ish video highlight roll off, but I think this occurs only when there is an accurate white balance to the light sources. (And it seems it wasn't properly white balanced in the above tests given the overall yellow tint of the scene)

    It also seems to be a fact that cinema retains more details in the midtones than the other modes (which is something else to consider in selecting the film mode you personally like).
  • On the matter of distinguishing preference and fact, a post from awhile back addressed it best:

    "I've only used smooth -2, since the beginning. I think you may be referring to this(link) post. I was responding to qwerty123, who was confident with his understanding of cinema. However, a few posts previous I wrote:
    "The hacks do improve the banding significantly. Use smooth -2 for top quality roll off'

    The situation is this; if cinema, whether because of the name, the in cam curve look, or the loss of the noisiest part of the DR, gives him more confidence to tell a better story, he should use it. When I was getting started in making movies with friends, I had no idea what color grading was, nor did I have access to the tools to do it. What the camera recorded to VHS is what we got. At that time, the more cinematic the image was on the screen, the more confident I was to tell a good story. Here cinema mode can be a godsend.

    For those gearing up to shoot a feature that might end up on the big screen, cinema film mode has no place. I assume that a skilled colorist will be working with the footage, if not by the original team, then by the studio who buys it and will pay a colorist to present in the best light. In this case, cinema is a huge disadvantage, because the bottom noisy part of the dr is lost and the rest is next to black (which is not a good thing even with hacked avchd, banding) and the roll off to highlight is allowed to be rendered after the blue channel blows out (which causes a hard-to-remove yellow shift on overexposed areas). Smooth is a big advantage, as it lifts the shadow (with entire dr, noise & all) away from pure black, and applies a bit of logarithmic curve to roll off the highlight without shifting to yellow. It also retains accurate color. For a good colorist, it is a few clicks to make it look better than cinema mode ever could."
  • @Sage Thanks for the response. I think I just disagree with some of your alleged factual statements: that smooth is more color accurate than cinema; that cinema per se (irrespective of white balance etc) causes a yellow roll off in the highlights.

    Although I do *agree* that cinema has less DR than smooth (about 2/3 stop less according Dpreview, if I remember correctly) AND that it looses more to the blacks than smooth. But blacks can usually be recovered by a basic color correction.

    I remember doing a test once in which I shot smooth, nostalgic and cinema of the same subject. I could easily make cinema look like the other two (raising the blacks, etc) and, I think, vice versa. The color rendition is a different story, and would require a lot more work.

    Also note that highlight roll off is usually not the place in your image where the most important subject will be (assuming that you are shooting human beings, etc). Cinema gives more of its DR to the midtones than smooth which has a bit more even distribution. It is also definitely worth comparing skin tones between the various modes -- I'm unsure anyone has a public test about that.

    I would stay away from cinema like the plaque if it did produce the ugly yellow roll-off, but, at least in my uses, I haven't seen that with proper white balance. I've done severe color correction with both modes (and others have as well), and both cinema and smooth hold up quite well. I personally like the color and detail rendition of cinema better to the extent that I'm willing to give up a 2/3 stop DR, but to each his/her own.

    Anyway, I think we're both getting off the topic of this thread (and I apologize for that).
  • After checking out these grabs, I had a formed a curiosity about how my GH2 with Quantum v2 hack would fair against my XF300. So here are the comparison grabs. GH2 had sharpness at 0, iso 320. XF300 was at -4 and gain at 0db. Used a Canon FD 85mm f/1.8 lens at f5.6 on the GH2 shot.

    These are 100% center crops, and the curve of the GH2 grab was adjusted to more closely match the XF300.

    I think it's pretty obvious that the GH2 doesn't resolve as much as the XF300, but I didn't expect it to since the XF300 is a 3 chip camcorder that resolves more than 1000 TVL. The GH2 does very, very well, though in terms of sheer resolution. I'm actually very impressed - to my eye it looks to be around the 800 TVI range.
    XF300 Sharpness crop.jpg
    831 x 410 - 512K
    GH2 Sharpness crop.jpg
    834 x 409 - 438K
  • @cowpunk52 "I think it's pretty obvious that the GH2 doesn't resolve as much as the XF300, but I didn't expect it to since the XF300 is a 3 chip camcorder that resolves 1000 TVI."

    uh oh, here it goes again. lol.

    Seriously, I'm not a fanboy... but this is just miss-information again. The EX3 is a 1000line camera, and my stills (and clips I haven't posted yet) clearly show it out-reolving the EX3 by a bit. Which means that the GH2 might actually be a true 1080p image with it's advanced down sampling.

    I also use the XF300 all the time and it's not even close to the GH2's resolving power in terms of fine detail. Not to mention the the edges are very digital and grimy. I just don't believe you can get good clean detail without over-sampling. Native 1080p images just don't look good. They look to digital. The only 1080p camcorder that looks halfway decent next to the GH2 is probably, well, the EX3...

    Again, I'm not trying to prove the GH2 is the best camera or anything just for the sake of it... I just honestly, haven't seen an image that betters it besides Red and the Alexa. Camcorders are just not even close.

    "to my eye it looks to be around the 800 TVI range"

    Look at my full-res stills again... the EX3 is 1000TVI. The GH2 has more detail. Therefore, it's more than 1000 TVI. To say that it now has only 800 TVI is illogical and going backwards now. It doesn't make any sense.

    ...unless it's your personal opinion or something. Then you can believe that it only has 500 TVI line with 6 stops of dynamic range if you like. Won't make it true though. :)

    @Sage

    Thanks for your thoughts on cinema mode. I've also had a feeling that the DR is not as good as smooth. It seems like the highlights handle better with cinema initially, but I think it just causes them to clip too fast. It just doesn't expose as faithfully as the other modes.

    I may have to do a "Round 2" with smooth vs the EX3 after the holidays...
  • @bwhitz - hey, man, I was just curious and did my own test with a different camera to see the results for myself. I didn't say a word about the EX3 or your test, only my own. And I posted the results, right here, two posts up, for everybody to see. From that test, it's clear the XF300 resolves more real detail than the GH2 does. I don't know why your test with the EX3 didn't show the same thing, but I'm not really concerned about it.

    I love my GH2. It's awesome, I use it all the time. But from my test, an XF300 resolves more fine detail than a hacked GH2. How else do you explain the results?

    But I just thought of another question - since the EX3 and XF300 are camcorders that have been tested and proven to resolve 1000 TVL, how could the GH2 resolve more than that? The physics don't add up there.
  • Does anyone know the technical reason for why the GH2 + street lamps have that secondary halo around them? Other DSLRs don't do it and I can never seem to change that behaviour no matter what the settings.
  • @cowpunk "But I just thought of another question - since the EX3 and XF300 are camcorders that have been tested and proven to resolve 1000 TVL, how could the GH2 resolve more than that? The physics don't add up there."

    GH2 has a 16mp sensor... roughly 5k. The Venus engine takes all of that 5k data and down-samples it (properly) to make a fully resolved 1080p image. The EX3 and XF300 are both native 1080p images. They don't benefit from oversampling.

    Basically, the GH2 is starting with MUCH more information and detail... and using it well. Why would you think that a native 1080p chip, starting with much less, would have an advantage here?

    Downsampling, when done correctly, always has more detail in the final image. Not to mention, the detail also looks more natural and the edges don't suffer from that digital look.

    "I love my GH2. It's awesome, I use it all the time. But from my test, an XF300 resolves more fine detail than a hacked GH2. How else do you explain the results?"

    Well, first of all... thanks for testing. Research and samples are always appreciated. But, personally, I can't determine much from the images you've uploaded. Maybe shoot some of the classic test subjects like trees, water, brick patterns, junk drawers, carpet... ect. ???