Personal View site logo
Make sure to join PV on Telegram or Facebook! Perfect to keep up with community on your smartphone.
GH2 720p50/60 and 1080i60 Settings
  • 254 Replies sorted by
  • @nomad

    @duartix is right, "/" means "or".

    However the main ProRes advantage over AIC is 4:2:2 chroma-sub, but GH2 support only 4:2:0 so I really don't know whether going for ProRes would be any better.

  • AIC is an I-frame only version of MPEG-2. Not really optimized for space, speed and generational loss. Plus, once you do any grading, filtering or noise reduction to 4:2:0 footage, you can generate intermediate values which profit from 4:2:2 and 10 bit.

  • I manipulate photos and images "professionally" (it's part of my job), so trust me on this: real input values can make a difference, not "generated" ones :-)

  • @daihard and @rikyxxx Thank you guys. It's working much better than Compressor+Cinema Tools

  • Settings used - Driftwood Quantum v9c 720@50p in all shots. Some shots was heavely colorgraded. This is my first try of Driftwood Quantum v9c patch and I very pleased with a quality of it. No record drop during all day long shooting session. http://player.vimeo.com/video/36882750

  • @Ralph_B. Any concerns with using Sanity 3.1 patch with firmware v1.1 and ptool v364d?

  • @Zaven13 It should work fine. BTW, I'm working on an updated version of Sanity with increased stability and spanning.

  • Use word settings insted patch like VK say.

  • I tried the patch with the new ptool. 720P look good, but the 1080i is only giving me 25 frames. Anyone else tried it?

  • Ralph_B. I used Sanity 1080i60 and 720p60 settings with firmware v1.1 and ptool 364d. I did a comparison testing of the same scene with similar shots between the old and new firmware, I get half the bit rates with the new firmware compare to the old one and the video quality is not as good. looking forward to an updated version of your Sanity patch with the new firmware and ptool.

  • *** NEW *** Vitaliy has just released an updated version of the recently released 3.64D dated 20.02 available from http://www.gh1-hack.info/ptool3d.zip.

    This updated version of the originally released 18.02.12 ptools 3.64D, incorporates 'switch on and record' 720p modes WITH B frames allowing you the b-frame prediction method which previously proved troublesome - even for Panasonic to implement!

    This update allows you to switch on B frames by selecting the Encoder Setting 1 720p = 3 found under the 'AVCHD Movie Mode/External Encoder Settings patch and utilise the B frames option already found in 1080p24 and 1080i recording modes.

    In the past the only way you could record with B frames on this setting was to record first under a different REC MODE like 1080p24 or 1080i for a few seconds and then switch to 720p mode and record. If you didn't do this you were met with an immediate record failure. Vitaliy's workaround now gets round this 'little technical hiccup' with some new code.

    NOTE: Encoder Setting 1 controls GOP frame parameters. Selecting any of the Encoder Setting 1 settings to = 1 ensures i frames for a GOP setting of 1. Setting this to =2 ensures only i and p frames are used in the GOP (GOP must be => 2), and finally setting it to =3 establishes i, p and b frames are used (GOP >3).

    I will be releasing shortly a STOCK GOP settings which demonstrates the new patch which will work on the GH2 and GF2. NOTE: This setting is a demo only of the B frames patch and can be fine tuned to work hi detail (like upping the IQ and Q slightly more until it doesnt fail on death charts, e.g. IQ =26, 1080 Q = 22 or 24 and 720 Q = 18 or 20)

    You can expect Chris Brandin to be working on an updated version of his 66M settings with scaled matrices and further options soon no doubt.

    ptools_200212_graphic.png
    378 x 649 - 35K
    Driftwood 66M- 35M Stock GOP all with B frames v1 - setd.zip
    630B
  • All of this is great fun!

  • Here's an updated version of "Sanity" (by Ralph_B) for PTools 3.64 / FW 1.1. Especially interessting for those who want to use HBR mode.

    I've tested Driftwood's Quantum X and LPowell's FlowMotion … Although the bitrate is much lower Sanity clearly looks better in HBR mode. In Quantum X and FlowMotion HBR looks more like in their respective FSH mode. But in this Sanity version 24H and HBR look almost identical. First in FSH mode you'll see some degradation (naturally).

    Basics:

    24H = 66Mbits | HBR/FSH = 36Mbits | GOP12 | AQ4

    All other settings are from the original Sanity hack.

    Attached a screenshot from a comparision Sanity and Flowmotion both at 24H and HBR. Camera settings: Smooth, -2/-2/-1/-2. ISO400. Lens = Zeiss 2.0/50. Workflow: MTS files converted through 5DtoRGB. Imported into Avid (DNxHD 10bit). Transcoded to x264.

    Here's also a link to the comparision movie (4'' uncorrected / 4'' blacks lifted at 16bit rendering): http://www.file-upload.net/download-4132670/flomo_vs_sanity_x264.zip.html (26MB)

    sanity_flomo.jpg
    1920 x 1080 - 727K
    seta.zip
    668B
  • Nice work, towi. Sanity definitely has less noise, or at least finer noise. Getting clean noise is one of the things I obsess about when I work on these patches.

  • @towi - "Attached a screenshot from a comparision Sanity and Flowmotion both at 24H and HBR... Workflow: MTS files converted through 5DtoRGB."

    So this is a comparison of 5DtoRGB versus 5DtoRGB? I think I'll need some help with this evaluation. I'm peeping, but I'm not seeing the pixel you're referring to "that clearly looks better"?

  • Ralph ... yes, the fine (or less coarse) grain is one thing. Regarding the noise pattern Sanity is on par with Dritwood's super highbit hacks at 24H. The other thing is how HBR handels the field integration (if this term even exists :-/ ). The fields/lines simply look clean and sharp in Sanity whereas in other patches it looks like adaptive deinterlacing has taken place for some reason.

    Many thanks for this great hack by the way. It's my standard patch as it offers clearly the best 50i quality and at the same time really great 24H. Now with FW 1.1 it offers the best HBR quality... IMHO.

  • @LPowell I admit it's subtle, but when you play the movie, look in the orangish area at top left, and when he lifts the levels, look at the black area in the center.

  • "So this is a comparison of 5DtoRGB versus 5DtoRGB?"

    No. 5DtoRGB was used to convert all the 4 files from MTS to MOV. That's all.

    "I think I'll need some help with this evaluation. I'm peeping, but I'm not seeing the pixel you're referring to "that clearly looks better"?"

    At 24H both look pretty much the same. Too, Sanity 24H and Sanity HBR look pretty much the same. But at HBR Flowmotion is softer than Sanity. Actually I should have made a comparsion with wide angle shots. But in my test shots the light changed and it was windy. Hard to compare. So I decided to use this close shot.

  • @towi Pehaps I've gotten the wrong impression about 5DtoRGB. My understanding is that it filters the chroma channels to create a higher quality 10-bit rendition of the original file:

    http://nofilmschool.com/2011/08/top-quality-dslr-plugin-5dtorgb-version/

    I don't have a philosophical issue with this approach, but it would alter the original pixel data in a way that would make comparisons between patches questionable. I'm not sure what you mean in referring to Flow Motion as "softer" than Sanity, but smooth and flicker-free motion picture quality is what I'm aiming for. From that perspective, comparison of the static image quality of individual frames is insufficient. For me, it's the motion quality that counts and that's where the difference between Flow Motion's 3-frame GOP structure and Sanity's 12-frame GOP may be most apparent.

  • "I don't have a philosophical issue with this approach, but it would alter the original pixel data in a way that would make comparisons between patches questionable."

    hmh... but it alters the original pixel data of every clip in the same way. Too, this is what we do with MTS files - we convert them into editing friendly formats ... at least I can't imagine anyone editing MTS files natively. Files transcoded through 5DtoRGB and the same files directly imported into Avid look the same (regarding colors/gradation). So I don't think 5DtoRGB really alters anything. Things are different when you transcocde to ProsRes 444... but I've transcoded to the regular ProsRes HQ 422 codec. Anyway ...

    I agree on the comparision of static scenes. But motion is very hard to compare... at least it's very hard to make those comparisions available on a forum (web).

  • @towi - "...at least I can't imagine anyone editing MTS files natively."

    That is exactly what I do with every test clip that I analyze. A video that's been transcoded can no longer be used to evaluate the encoder that originally recorded it. At this point, the only practical way to compare footage is to download exact copies of the original files.

    "So I don't think 5DtoRGB really alters anything."

    Isn't that the whole point of using 5DtoRGB in the first place? Because it filters the images in a manner that you prefer to the original?

  • "Isn't that the whole point of using 5DtoRGB in the first place? Because it filters the images in a manner that you prefer to the original?"

    I don't care about the "original". Because you can't really edit MTS files in real time. Too, MTS files do not contain reel names and unique timecode. This is why I convert MTS files to QT MOVs in any case. So for me personally it's just a question of the best converter. So if you want so: I prefer 5DtoRGB over ClipWrap or any other converter (and also over direct import into Avid beacause I first assign reel names and timecode to my QT MOVs prior to import into Avid).

  • @towi: "I can't imagine anyone editing MTS files natively." "You can't really edit MTS files in real time."

    Many people who use Adobe Premiere Pro or Sony Vegas edit natively with MTS files. It works great for me.

  • I'm one of them. I've been editing music videos using native GH2 files without issue on Premiere Pro.

  • @towi - "I don't care about the "original".

    That's fine, but it means that your qualitative comparisons are not of Flow Motion or Sanity, but are instead esthetic evaluations of the products of your entire workflow. While that's a perfectly reasonable way to produce high quality results, it's not an unadulterated test of encoder image quality.