Personal View site logo
Make sure to join PV on Telegram or Facebook! Perfect to keep up with community on your smartphone.
Please, support PV!
It allows to keep PV going, with more focus towards AI, but keeping be one of the few truly independent places.
GH2 High Quality GOP3 Settings, Series 2: 'AN' Valkyrie, GOLGOP3-13, GOP3ZILLA, Duke, natural :-)
  • 670 Replies sorted by
  • @driftwood Thank YOU Nick! :-)

    I am going to improve Spizz in addition to 24p more.

  • @bkmcwd I'm just curious, do you think the new Spizz will do a better quality than Duke or Valkyrie? Btw, I agree with Towi, Duke (Beta1) produces a very good quality in PAL, I will also test Beta 4 if I have a bit more time.

  • @spreeni

    Hi mate! :-)

    do you think the new Spizz will do a better quality than Duke or Valkyrie?

    This is a matter of taste. Personally, I prefer Duke or Valkyrie. However, I interesting challenges in Spizz.

    Duke (Beta1) produces a very good quality in PAL,

    Supposing you can use it satisfactorily, to be sure, BETA1 will be very good. However, there is a risk of freezing in it strictly. I feel appropriate bit rate of 25p in PAL as being around 60M.

    Now, I am repeating the experiment in order to further stabilize. The result is due to be harnessed in Spizz or other settings.

  • bckcwd, sorry it has taken me a while to reply, haven't been on!

    I used the version released on 07/18/2013, a few posts above my original post.

    I used a SanDisk Extreme Pro 95MB/s 32gig

    I was in manual movie mode.

    And the shot was a wide shot.

    Hope this helps, and thanks again :)

  • @rahorto

    Many thanks for your feedback! :-)

    SanDisk Extreme Pro 95MB/s 32gig

    Also at the same speed, the difference between SDHC and SDXC is more than imagination. Even so, now, I have to stabilize it further.

    I seem to be able to release the new version soon.

    Thanks again.

  • Hi,

    Since @towi made the wonderful test movie, I introduce it. Although it is difficult to clear this with iDRC in 60p, I am adjusting in principle so that all the settings of mine can clear this. :-)

    *When using this, it is effective to make PC monitor the brightest and to try this on around ISO1000.

    Many thanks towi!

  • Valkyrie 444 TYPE-ZERO4 & Lite BETA4.3

    bkmcwd "Valkyrie 444 TYPE-ZERO4 BETA4.3" (Standard) - seta.ini - Released 12/08/2013

    bkmcwd "Valkyrie 444 TYPE-ZERO4 Lite BETA4.3" - setb.ini - Released 13/08/2013

    *All the rec modes were changed from BETA4.2. :-)


    This TYPE-ZERO4 is also the version as which "balance of gradation and resolving" was considered in all the modes. (Scaling Matrices are the same as TZ3 or previous Lite.)

    I think that the freezing issue at the moment of the focusing and zooming was solved nearly perfectly also in Panny lens use.

    Although towi's Death Movie is cleared in all the modes, there is no telling whether span is possible in other than 24p.

    Cautions are required to use iDRC or iA only in 60p.

    Please try it! And I am very glad if I can get feedbacks from you.

    *If this is also still unstable, I will further adjust.

    Special thanks to @Azo, @towi and all silent users of my settings! :-)

    Of course, many thanks to @cbrandin and KING @driftwood!

    • Superb-looking Pure Original 444 matrix from cbrandin is adopted in 24p mode

    • Modified 444 matrix from my BETA Team in Japan is adopted in HBR/1080i and 720p modes

    • Upto 140M for Top Quality 1080p24 at Q18 / 'Standard'

    • Upto 140M for Top Quality 1080p30 at Q18 in 24L mode with 80% Slow-Motion Mode (No Sound) / 'Standard'

    • Upto 110M for Top Quality 1080p24 at Q18 / 'Lite'

    • Upto 140M(110x1.25=137.5M) for Top Quality 1080p30 at Q18 in 24H mode with 80% Slow-Motion Mode (No Sound) / 'Lite'

    • Upto 75M for Top Quality HBR/FSH modes

    • Upto 75M for Top Quality 720/50p modes

    • Upto 75M for Top Quality 720/60p modes

    • Extremely reliable top performance in AVCHD 24H and 24L

    • 80% Slow-Motion is available in 24L only with 'Standard'

    • 80% Slow-Motion is available in 24H and 24L both with 'Lite'

    • Fully working iA and iDynamic range modes if you require them (Take care only in 720/60p)

    • Beautiful 3 GOP bringing short GOP style detail & movement-rendition in all the modes

    • Bush of Death / Death chart tested.

    • Wonderful Pasadena Pulse Audio Patch V2 B2, by Per Lichtman

    • Top quality 2K mjpeg Mode in HD by Nick Driftwood

    • In camera playback is sometimes not available in 24p. (Standard) In other mode, Sometimes available after reboot the camera.

    *Supposing you use "SH" mode safely in 60p, please turn OFF iA and i dynamic range control. On the other hand, in other rec modes, it is satisfactory at all.

    *I especially recommend you use with SanDisk 95MB/s 64GB SDXC card.

    bkmcwd Valkyrie 444 TYPE-ZERO4 BETA4.3.zip
    1K
    bkmcwd Valkyrie 444 TYPE-ZERO4 Lite BETA4.3.zip
    1K
  • I will touch Spizz anew, after releasing Valkyrie Lite after this. :-)

  • @bkmcwd

    Katsu San thank you for updating the Valkyrie 444 TZ4 patch :) So I am assuming that the @towi death screen movie puts more stress on the camera then the old death screen movie is this correct? If so I will run the tests on all of the patches that I plan on using. Thanks @towi for the new death screen....

  • @Azo

    I am assuming that the towi death screen movie puts more stress on the camera then the old death screen movie is this correct?

    yes, much more stress. Driftwood, bkmcwd and me are using it now as we need a common denominator.

  • @Azo @towi

    I am assuming that the @towi death screen movie puts more stress on the camera then the old death screen movie is this correct?

    It is also yes and also no. It is because the setting which cleared this Death Movie is also often frozen by static Death Screen etc.

    I have been doing at least four kinds of tests recently.

    1. static Death Screen

    2. moving Death Screen on static Death Screen combination

    3. towi's Death Movie

    4. continuation zooming on static Death Screen

    All these are tried on max brightest monitor with ISO1000/SS800 etc.

    Since these tests will try the element with which each is different, I cannot say at a word which brings about the highest stress. :-)

  • @bkmcwd

    Have you performed this extensive suite of tests on your latest Valkyrie 444 TZ4 patch? This would be good to know because if you have it shows that the patch is really stable and solid.

    非常にカツサンありがとう

  • @Azo

    Have you performed this extensive suite of tests on your latest Valkyrie 444 TZ4 patch?

    Of course! :-)

    However, it still may not be perfect. It is because I did not try all the lenses in the world. Although I think that it is probably OK ...

  • @bkmcwd

    However, it still may not be perfect. It is because I did not try all the lenses in the world

    LOL :) that was funny and yes I understand there are to many variables in terms of subject matter, lenses, etc. etc. Anyway glad to hear that you put Valkyrie 444 TZ4 through the ringer. I will try it out today at the Beach :)

  • @bkmcwd

    the setting which cleared this Death Movie is also often frozen by static Death Screen

    in my experience when the codec survives the first 34 seconds (including the black frames / black wipes which push the frame size to the max) it will also survive any static torture tests. Too, when I am testing new settings I always start with recording 2 minutes of the first frame of my codec torture movie (so the wired B/W chart) and check the Streamparser data. First when everything went fine I am checking the setting with motion.

    Then again, as you know, I never use auto-anything... so when a codec survives my torture movie it will (most likely) survive any real world scene ...

  • UPDATE(2013/08/13):I have released Valkyrie 444 TYPE-ZERO4 Lite BETA4.3 http://www.personal-view.com/talks/discussion/comment/137212#Comment_137212

  • @towi

    Then again, as you know, I never use auto-anything... so when a codec survives my torture movie it will (most likely) survive any real world scene ...

    Yes, this is the difference of a position between you and me. :-)

    For example, if WB is changed, a result will change considerably. However, it is physically impossible to check all the WBs. Therefore, I always use AWB for the test.

    There are many such other elements. This is not for using auto but is a method for making it stabilized in every setup.

    On the other hand, I also think that free arrangements are attained in the settings used only by limited setup personally.

    I attach greater importance to Buffer Analyzer than to Stream Parser, when testing in these DeathScreens.

    Valkyrie444_TZ4_BETA4.3_24p_ISO1000_towisDeathMovie.jpg
    1073 x 932 - 525K
    Valkyrie444_TZ4_BETA4.3_24p_ISO1000_continuouszooming.jpg
    1071 x 933 - 527K
  • @bkmcwd @Towi @Driftwood and co

    Why not making a unified patch ?

    @Towii

    Torture movie doesn't represent real world scene, it is more complex than that... but the tortue movie is a good way to test the stability (nothing is perfect) so you will always find a way to improve the execution

    I was thinking to make my own patch based on my lut algorithm (ARRI and co) with the lut generator help http://www.arri.com/camera/digital_cameras/tools/lut_generator/lut_generator.html

    But i don't have time for it.

    From what i saw on ever patch from Drift/bkmcwd/towi, the final result look closer on each patch you made, probably because you keep the same typical render, it is not a critic ! btw thanks for your work, it's just that you should maybe think about looking another spot to give a little bit flawless feeling (new render).

    A tip you probably know, for the white keep the sensor (without lens) on a white wall (not yellow, white) and record few second of it, check the white color information and improve it (use real object surface, not video). Same for the black color (you can leave the sensor cover on it for that). Is actually one of the best tips to read what the sensor take, take every surface/object as a example to improve your picture (i mean use color surface from a real scene)

    P.S. i guess i need to make one too... this is what i reach with the GH2, keeping a strong dynamic range, flat and detailled picture (well i think, we don't have the same perception)

    (the uploaded picture is a capture video screen)

    Test.jpg
    1024 x 640 - 932K
  • @zilk: Which patch exactly is used for this capture video screen?

    Which GH2 profile and parameters did you use to get such a ultra flat image? Which lens?

  • I don't use any patch, it's a old algorithm work that i needed to improve but didn't have time to finish. i have others stuff to do, smooth hack profile + new color algo/spectrum and extra stuff that a high end users can't touch it (not a good idea).

    Lens used was a 17mm 2.8 Olympus.

  • @bkmcwd I made a quality test of duke beta4 yesterday. I only testet my prefered modes (PAL) FSH and SH. The quality is much better than the actual Spizz (T6 preview) or former GOLGOP in my eyes.

    Driftwood compared Spizz with FlowMotion and told us, Spizz wins the streamparser analysis game. But to be honest, I cannot see a quality improvement from Spizz T6 to Flowmotion in real life. In test settings with same lightning FM shows more texture the dark areas than Spizz. Duke handles these parts better, too. So for me, Spizz is out of the race. As I wrote, I only looked at PAL FSH and SH, maybe NTSC side or 24P looks different.

    One question to Duke, did you use a different matrix than the last GOLGOP?

  • @spreeni

    Driftwood compared Spizz with FlowMotion and told us, Spizz wins the streamparser analysis game. But to be honest, I cannot see a quality improvement from Spizz T6 to Flowmotion in real life

    No, he did not... he was referring to 24H only! All other modes are not yet ready for primetime. But personally I believe in the end it will be a very, very good setting.

    For PAL modes you may also take a look at "HQ3GOP": http://www.personal-view.com/talks/discussion/comment/137327#Comment_137327

  • @spreeni

    I've just finished my part on Spizz T6. @driftwood and @bkmcwd have to verify it now...

    Here's a little "preview" of HBR/25p

    • 1st three attachments: Codec Torture Movie

    • 4th attachment: low key scene

    • 5th attachment: medium detailed scene (underexposed)

    spizz_t6_towi_HBR25p_codectorture_Frames.PNG
    746 x 412 - 44K
    spizz_t6_towi_HBR25p_codectorture_Frames2.PNG
    746 x 412 - 45K
    spizz_t6_towi_HBR25p_codectorture_Frames3.PNG
    746 x 412 - 48K
    Spizz T6 towi HBR25p Low Key.jpg
    1201 x 812 - 490K
    Spizz T6 towi HBR25p Texture Medium Detail.jpg
    1233 x 808 - 503K
  • @zilk

    Why not making a unified patch ?

    Because, there are several reasons.

    This is the biggest reason in the first place, GH2 hack has not yet been elucidated all still. And there is a reason the picture of an ideal changes with people.

    Anyway, we are still in the middle of a way. Based on the feedback from various users, it is the stage which I am also always improving. :-)

  • @spreeni

    Thanks for comment, mate! :-)

    One question to Duke, did you use a different matrix than the last GOLGOP?

    Yes, they are different. In Duke, resolution is more important than GOLGOP3-13s.

Start New Topic

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Sign In with Google Sign In with OpenID

Sign In Register as New User

Tags in Topic

Top Posters