Personal View site logo
SANITY 4 and 5.1
  • 1139 Replies sorted by
  • @bmcent1 Thanks for posting. That isn't the least bit subtle, is it? I haven't seen that on 5 yet but haven't taken anything with that dark a background either.

  • Yeah, I'm pretty new to video and particularly to the GH2 so I don't know exactly what to expect, but it did surprise me. The brighter exposed shots looked really nice, but as soon as I closed the aperture down, the banding (if that's the right term) was pretty clear.

  • @bmcent1 When you say banding, I assume you're refering to the dark horizontal bands. This has nothing to do with Sanity. This is typical of the GH2. It does this under certain conditions. You'll even see it in the uncompressed HDMI output. BTW, that's a mean looking mother...

  • bmcent1 it's best to use ISO increments at the top, 160, 320, 640, etc.. so 200 ISO would be worse than 32O.

  • Yes @Ralph_B, that's what I was referring too. In a way, sorry to hear it's the GH2 and not the settings ;) I know the scene was darker due to my aperture, but I was outside, middle of the day, in shade. I guess the only way to avoid those darker horizontal likes is to make a better exposure then darken in post if desired?

    Funny, it he is mean looking, but it's mega magnification with the 150mm and ETC crop ... he's actually smaller than the lead point of a pencil and easy to miss with the naked eye!

    Thank you islander, I will stick to those then!

  • @Ralph_B That's interesting about the dark band I wonder if that's what I've taken as Vignetting in some of my lenses.

    @islanders66 I thought the middle numbers were the quick way around the ISO bug?

  • @peternap, yes to avoid ISO bug navigate with middle numbers from highest to lowest then go up to the top row. I just attended a workshop with DP for SNL and he explained the importance of using the increments of 160, 320, 640 etc. Kind of like using double the frame rate for shutter speed.

  • bmcent1, this is the infamous GH2 dark band that happens at high ISOs and apparently also with ETC mode. The problem is minimized by using a longer exposure time.

    islanders66, your advice about ISOs is not correct for the GH2. It's more complicated and subtle than that.

  • @blazer.. how is it not correct? Arent the ISO on the top row all clean if you avoid the bug?

    " ISO 640 is a bit less noisy than ISO 500 if you're stepping down from 800 (avoiding the ISO bug), and a little bit noisier than 800 if you're stepping up from 500 (invoking the ISO bug). But ISO 640 is less noisy than 1000 and more noisy than 400, no matter what."

    So you are saying that for some reason 400 is also a sweet spot? But that doesn't mean you want to avoid all the ISO on the top row as a short cut to avoid the bug, as this was the context I was responding to. In fact you want to use ones on the top row because they are all clean. I can't remember exactly why, and my source could be inacurate for GH2 as blazer claims, but as I recall it had something to do with the number of pixels that translates into an even distribution for these increments, which is why they are on the top row?

    I would like to know why this doesn't apply to the GH2? If it is too complicated for someone to use the top row, how will they be able to remember exactly which other single ISO is ok while avoiding others on the same row? For me and my workflow it's easy to just use the top row and avoid the ISO bug.

    I haven't noticed any banding issues after I got a fast lens and use double the correct framerate.

    I notice lots of noise with a slow lens.

    These are just my observations.

  • Thanks for the link @balazer but I don't think my is the FPN discussed in that thread. Mine jumps all over the frame, is different thicknesses from frame to frame, and isn't in every frame in the clip. In some frames there is only one band, in others there are 2 or 3. It's very much like noise but instead of being individual noisy pixels it's full lines that are under exposed or have less red in them. In the thread you linked, they were referring mostly to a constant underexposed band which I didn't see in my clip yesterday.

    To me, it looks like it must be some sort of amp gain or read/scanning variation across the sensor. But I wondered if it had anything to do with Sanity because I wondered if the codec or tables (don't know how all that works) would introduce this look if there was a very high or low reading in a line that had very little detail because most or all of the line is out of focus and shadow. In one of the images I posted the band seemed to line up with the detail on the spider's leg. But maybe that was just a coincidence.

  • Well, film the same scene with the same settings first with Sanity 5 and then with the standard firmware and compare the clips. Then you know, if it is a Sanity or a GH2 problem.

  • I would suggest testing it with the ISO on the top row and see if that helps. Come to think of it I had some banding issues when using the ISO on the other rows. Not sure exactly what it was but haven't noticed anything like that since with a fast lens, top ISO and correct shutter.

  • @RalphB said: " I discovered a way to create monster size I frames, while still keeping the bitrate low."

    Cool!!! So what's the trick? The scaling tables, Auto Quantizer or both?

  • islanders66, you can pick up the ISO noise discussion in the other topic.

  • @duartix I vouch for the monster size I frames (manipulating scaling tables for P&B frames I believe), however it is my opinion that this creates artifacts in darker areas of the image. I will have to test different scaling tables to confirm this.

  • @pdlumina Load the footage in StreamEye. It's easy to see if this is related to the quantizers.

  • I don't use shareware.. plus I'm in Ubuntu :) I wonder if the GOP size affects the details. 17 frames is a long way from I frame to I frame, particularly with movement.

  • It's not Sanity, I guess it is the sensor or maybe AVCHD (because of what I noticed about MJPEG below...)

    I shot the same scene today with Sedna C and it's doing a very similar thing with pulsing, full frame width, thick horizontal lines at random vertical locations from frame to frame. I'm shooting at ISO 160 (after going to 200 first) in ETC mode for both Sanity 5 and Sedna C Q20. Both look great outside of ETC and I only see it as I close the aperture down, ie not at a well exposed shot, and also not when the scene is very dark, but as it gets darker while recording a big highly defocused background.

    Interestingly, I don't see it in MJPEG HD in Sedna C Q20 even in ETC mode. I didn't think to test MPEG HD in Sanity 5 before swapping the firmware.

  • delete my post, wrong settings.

  • Sanity 5. Looks promising, i like the way it allows smooth movements (reduced jitter while panning), great, well done ... Will do some test's also would be nice to have a GF2 version too! :-)

  • @Ralph_B

    I've quickly tested your new settings.

    in HBR/25p mode I-frames hardly exceed 600K. Maybe just numbers... but infact the video looks compressed. Sanity 3 rocks in HBR/25 mode...

    Your concept of using the fallback scaling tables for P- and B-frames is very interessting. It just doesn't work for personally me at 25p. But it have given me ideas for other settings. So many thanks for bringing this up!

  • So I shot some Sanity 4.1 today. Maybe I'm missing something, but all the HBR files were 30p? Not sure how to get around this...

  • Sanity 4.1, very impressed

    Screen shot 2012-04-19 at 1.14.25 AM (2).png
    1724 x 1077 - 2M
  • Aside from slight sky banding, wow!

  • Sanity 5 hack 35-85mm f2.8 zoom Vivitar Series I Lenses (of 1980) Vibrant -2 0 0 -2

    the grain you see appeared after it was sharpened