Personal View site logo
Official Low GOP topic, series 4
  • 586 Replies sorted by
  • @artiswar, glad your liking the lomos ;)

  • @shian - That is 9b.

  • @artiswar which patch did u test in those screen caps? 9b?

  • Well I started the feature on Aquamotion v2 and for consistency will continue to shoot that patch... but next film--- which is a shorter film (1 weekend shoot). I think I am going try out Quantum X...

    *In Driftwood I trust!

  • Will Quantum X work with panny lenses... a conformations would be great all i need is 24H

  • @driftwood Have you gotten the i frames higher on 50/60? (~280?) If so, count me in

  • @Driftwood Cheers Nick.

    Reading @balazer last comment, what do you reckon is the magic parameter setting that makes a static quantization work? I'd kill for something as simple as this:

    h264qc.png
    875 x 709 - 103K
  • Quantum X - the all i frame patch is coming...

  • @JDN The one on dvx is more recent. I want to set the standard to your test, so that it works for you 100% of the time.

  • @duartix, to achieve constant quality in my Cake patch settings, I set the bit rate and the frame limits high, and set a quantization parameter to control the quality level. I copied Driftwood's Quantum Beta 8, which apparently has some other setting in it necessary to make a static quantization parameter work.

  • @balazer:

    What are the essential parameters that make this patch set so much VBR?

    What other parameters did you have to watch for it not to break?

    I'm trying to make a patch set (for totally different footage characteristics) for which I believe this VBR approach is the way to go and I'm trying to learn how to tackle it.

  • It seems that lpowell is trying to steal some thunder from my Cake with his Flow Motion. ;)

    At first glance it looks like Flow Motion has a quality advantage if you don't care about spanning. If you care about spanning, Cake should win for quality and efficiency.

    But I need people to test it and let me know if it works!

  • @sage, yeah, it's the new nokton, 100 percent manual. It was the patch from the patch vault, not the dvxuser, so if that's a bit newer let me know and I can test that.

    It may have been what I was shooting -- millions of moving grains of sand plus turbulent water is pretty stressful on the codec. Alternately, it may have had something to do with the shutter speed (it was really high when it started failing as it was for twixtor, and to tell you the truth, I don't think I tested the nokton at a lower shutter speed -- just switched up to 9b as didn't have time to really mess around at the shoot (had only pretested the 50mm and thought since that manual worked all would... but didn't).

  • @Sage -(nine updated) ive had good results on 720 sh with canon lenses, did a bunch of records up to 30 seconds until i stopped recording because that suited my needs, detailed tree shots with movement

  • @balazer: A VBR constant quality approach? What kept you so long???

    What particular settings need to be watched out for?

  • Do you love the quality of high bit rate intra-frame patch settings, but not the huge file sizes and short recording times? Announcing my experimental patch settings for 1080/24p and Variable Movie Mode, "Cake": inter-frame constant quality at a variable 20-80 Mbps, with SPANNING using SanDisk Extreme HD Video cards. A new approach to GH2 encoder configuration yields consistently high quality in every frame, while using predicted frames for greater efficiency and high bit rates only when the video demands them.

    These settings are an improvement over existing long-GOP settings, offering better quality and efficiency. The quality approaches that of some high bit rate intra-frame encoder settings like Aquamotion.

    Here are sample bit rates for a static scene shot at different ISO settings:

    ISO 200: 23 Mbps

    ISO 400: 29 Mbps

    ISO 800: 43 Mbps

    ISO 1600: 57 Mbps

    ISO 3200: 66 Mbps

    ISO 6400: 73 Mbps

    ISO 12800: 77 Mbps

    Use these patch settings in 24H mode. In case of problems, I've set 24L to an untested lower bit rate for use as an emergency back-up. Feel free to tweak the 1080 quantization parameter: lower numbers yield higher quality. But don't go too low, or else the I-frames will bump up against the frame limit and look worse than the P-frames. A frame limit is necessary to allow spanning. These patch settings should work with any Class 10 card, but I haven't tested with other cards.

    Please report the results of your testing here. Special thanks to CBrandin and Driftwood, on whose work this is based.

    Balazer GH2 'Cake' v1.0 - seth.zip
    824B
  • Screen shot from a little test day. Going to edit some stuff together but I am STUNNED at the amount of detail. The first screen shot is set to "Fit" in Premiere. The second is at 400% the original size. On a 64gb 90Mb/s Sandisk card and a 28mm/f2.3 Lomo prime.

    Screen shot 2012-01-24 at 10.40.59 PM.png
    1920 x 1200 - 1M
    Screen shot 2012-01-24 at 10.41.14 PM.png
    1920 x 1200 - 333K
  • @sage @driftwood is quantum v9b 24h not able to work with panny lenses? Please verify

  • @JDN Nokton is non-electronic, no? I haven't been able to cause failure on my manual glass stopped down. Was it with the up to date version (on dvxuser)?

    Note: it *should fail instantly with the Panny lenses, as they have software distortion correction

    I am curious if another setting in the camera may be affecting stability due to higher cpu use.

  • lolwut

    Beast-Like

  • Quantum X is near completion. Nuff said.

  • Tested the quantum 9b and sage variant on a shoot today -- very high detail (grains of sand moving underwater while waves move above).

    Sage and 9b worked fine at 24p settings on sandisk 95mbps. 720p was mixed bag.

    On sage "NINE" setting, 720p60 caused write errors on all panny lenses as well as Nokton 12mm 1.6 almost instantly. Almost no problems on static shots with 50mm FD 1.4. 9b quantum held up a little better with panny lenses and the nokton, but still failed ocassionally. Near universal failure when trying to pan, but better on static shots. Didn't test spanning.

    ISO 640 with frame rates from 120 to 1000+ (doing some twixtor tests with the footage). Quality looks excellent on that which did record tho.

  • Such a childish overreaction to a seemingly innocuous question.

  • Shot these with Sanity... but not happy with my compression quality... its h.254 5k limit every 30 key frames? how are you guys getting such crispy quality ?? help

This topic is closed.
← All Discussions