Personal View site logo
Make sure to join PV on Telegram or Facebook! Perfect to keep up with community on your smartphone.
'Apocalypse Now' Experimental Series 1 Thread - BOOM, Intravenus - cbrandin/driftwood AN Soft/Cinema
  • 1089 Replies sorted by
  • I agree that screenshots aren't ideal because there is the issue of adjustments made to accommodate the display colorspace, etc... , even if screenshots are uncompressed. However, PNG or JPG files are even worse because the image has been recompressed. It's impossible to separate artifacts caused by H.264 and artifacts caused by the recompression. Vegas, for example can export as JPG or PNG, and the two don't look exactly the same. What is needed is an uncompressed format - like BMP.

  • @bkmcwd I used your settings "Valkyrie soft" for an interview on Tuesday. It was shot in HBR and ACCIDENTLY 50i on the second cam. Despite the wrong interlaced setting, the footage turned out nicely. I really like skintones!

    This morning I saw that you released another version called TYPEZERO1. How does this one compare to Valkyrie soft? Softer? Sharper? Or something else?

    I also read the following:

    "Warning! : in this setting, all the modes of AVCHD other than 24p are experimental. Please keep in mind that artifacts often appears in HBR, 1080i and 720p in spite of being very stable. These will be improved from now on."

    SInce I am interested in mainly 25p, are the HBR/50i settings of Valkyrie soft also experimental and may lead to artifacts? Or is this warning just directed at TYPEZERO1 testers?

    In our footage I didn't see any artifacts so far. I and was planning on keeping it in my cam for a an upcoming corporate shoot and hope it is not experimental. Don't want to find out in post later...

    Thanks again for the great settings!

  • @Tobsen

    Thanks for using Valkyrie and comment! :-)

    "How does this one compare to Valkyrie soft? Softer? Sharper?"

    I think that 444 TYPE-ZERO1 is "Softer" than Valkyrie 444 soft.

    And "Warning" is just directed at "444 TYPE-ZERO1" testers.

    "444 Soft matrix" is 444 matrix modified by Nick.

    Probably, it will be better not to use 444 TYPE-ZERO1, if you mainly use 25p. If it is 444 Soft, also in 25p, I think that it is satisfactory.

  • The sample of "Valkyrie 444 TYPE-ZERO1" with Pany lense. Please DL original MTS file to look the REAL IQ from vimeo.

    Pany 14-42mm F8.0 1/50 ISO320 NOS.(-2,+1,0,-2)

    Valkyrie_444_TYPE-ZERO1_Sunset.jpg
    1297 x 634 - 256K
  • @CBrandin PNG supports mathmatically (not just visually) lossless compression, but Vegas also supports TIFF output (and I seem to remember BMP) and VirtualDub supports BMP output as well. When I export PNGs from VirtualDub, they have been mathmatically lossless in the past. I will be happy to do an inversion test again vs the BMP output to see if that holds true in the current version.

    But PNG (except when used at lower color depths, like 256 colors) is setup for mathmatically lossless compression, not visually lossless. So any discrepancies would be caused by a mismatch between the encoder and decoders support of the format.

    I use PNG as opposed to BMP because the file sizes are smaller and the format is more widely supported in browsers for inline viewing. But I also pretty exclusively use it in 8-bit so I get lossless results. The compression ratios/quality dialogs for some programs can be confusing on that point but think of it the same way you do ZIP file compression: more compression means smaller files and longer to decompress, but it does not mean a loss of data.

    I completely agree with you about JPEG, though. That is why I usually post PNG.

  • @bkmcwd: the footage looks really nice. But in the lower left on the electricity lines there seem to be some artifacts...maybe my eye is fooling me, i dont know.

  • I really can't decide which I like more... :(

    SO THANKS GUYS! :)))))

  • @Mirrorkisser I also noticed. Possibly, although adjustment may be necessity slightly, I am already going to look at the result possibly more for the time being. This is a new starting point. :-)

  • @driftwood my initial tests of AN Boom Flat looks very good. No lock ups with rec or playback using SanDisk Extreme 30MB/s 32GB cards. Image looks very flat and skin tones look great, grades well in AE and AMC, nice fine grain at iso's up to 1000. The highlights seem to roll off very well with the picture profile set to smooth -2 all round and banding is much smoother than any other patches I have tried and it has good detail in the shadows with much less macro blocking. 24h mode cadence is smooth. The MTS files now have a more constant bit rate @ 142 mbps. I'm not concerned about card space or spanning for my broadcast work, it's more about getting the best I can out of this little beast. This patch makes my Canon lenses look even more creamier - which is what I like, and works well with m43 lenses. Thanks for all your tireless efforts.

    00081.MTS_snapshot_00.04_[2012.09.07_01.32.bmp
    1920 x 1080 - 8M
    00081.MTS_snapshot_00.04_[2012.09.07_01.33.png
    1920 x 1080 - 4M
  • @Ian_T I didn't shoot too much in low light, but from what I did I didn't have any noise issues at all.

  • So in theory, the Valkyrie patches will have more to work with in shadow areas yes? Or am I over-thinking again?

  • @MisterBink

    "the Valkyrie patches will have more to work with in shadow areas yes?"

    Although I am not sure on the basis of what you judge that it is "more", I consider yes. However, we need feedback more. :-)

  • @bkmcwd thans for the clarification :) I will test a bit more this weekend and most probably use it on the shoots next week ;)

  • @blaisejadoul I shot the interview using two cams and the Valkyrie 444 soft settings in HBR mode. I stopped the recording twice to be sure it wouldn't stop by itself. It turned out well!

  • @Tobsen How much time did you record ? And did it span ? I shot several tests with AN 6GOP soft in 24H and I stopped recording after 37 min and 48 min. Of course it spanned without any problem. I think that I'll stick to that for the moment...

  • @thepalalias - On my Vegas Pro the only options shown are PNG and JPG - they're both compressed and there are no save options (quality, etc....). Is there some option I have to set to get the other formats?

  • @Tobsen It's not just a question of spanning. It's also the fact that @cbrandin warned us that the soft matrices have been developed for 1080p !

  • @cbrandin, if i am not wrong it is PNG is lossless and 32-bit (includes an 8-bit alpha) in vegas.
    When you take a screen-capture, make sure preview is set to best, and the timeline is set to the correct frame format.
    Source: https://www.custcenter.com/app/answers/detail/a_id/711/kw/snapshot

    But you can "render as"
    Available options have been expanded beyond JPG and PNG to also include TIFF, BMP, DPX, EXR, and WMPhoto
    Source: http://www.videouniversity.com/articles/vegas-pro-9/

  • @cbrandin I don't know what you're using to play the video Chris, but http://sourceforge.net/projects/mpc-hc/ lets you grab a frame in png during play.

  • @mozes @cbrandin Glad you guys already cleared that up. But let us not go back to the old way of talking about color bit depths. :)

    In the mid-90s, we talked about color depth in terms of the aggregate bits from all the channels (for instance RGB= 8+8+8=24-bit or RGBA= 8+8+8+8 = 32-bit) but in the project properties for Vegas (and the discussions on this site) the standardized usage is bits per channel. So both the RGB and RGBA examples I wrote out would be 8-bit, as would be the PNG files I mentioned. :)

  • In Vegas, the PNG files are compressed. I'll look at other options and try to find something that produces uncompressed versions (or, some other uncompressed format). I have the Adobe stuff too, so I'll look at that as well.

    Thanks guys!

  • Since nobody who goes to the trouble of installing new settings is likely to use the footage out of the camera without color correction and other processing, maybe the issue is less what the footage looks like unprocessed, than what can be done with it.

    This would leave us with different questions: does one setting produce images with more obvious and uncorrectable defects than another, beyond which data rate are you encoding nothing useful on the GH2, and do certain settings produce footage which can endure more post processing than others?

    If the past is any guide, none of these question will ever be answered.

  • @jrd - good points... One of the goals for the "soft" matrices was to provide better latitude in post. There is a metric which is a pretty reliable indicator of how well footage will do in post. If you decode a clip using the JM-SP decoder in StreamParser it will show you something called "DC" value. The lower that number is, the better (below a value of 5, however, isn't necessary). The DC value is the lowest effective quantization parameter in the frame, and therefore an indicator of how well detail is rendered from a gradation standpoint (a value of 5 means that detail is rendered with 255 levels in the frequency domain - the maximum 8-bits can produce). Also the "Range" value is significant. Ranges of 5 or lower are fairly ideal. The lower the range, the better.

This topic is closed.
← All Discussions