Personal View site logo
Driftwood Settings, Series 5: Cluster v6 ULTRA RELIABLE, Mysteron, Crossfire, Quantum 9B, Sedna...
  • 1029 Replies sorted by
  • Shot mostly on Orion 4d. Paid gig. Big thanks to @Vitaliy_Kiselev @shian and @driftwood

  • Something (somewhat) narrative shot on the Sedna A with some production to it:

  • LOL, very good @kingmixer, great work, superb acting to excellent production all round. Congrats.

    @artiswar Really nice editing and great use of Orion. Well done.

  • Wow thanks! Awesome coming from you

  • @Driftwood I used Cluster V3 yesterday to video a Marathon. That's always an imprecise art because things are happening fast, it was bright sun and I had to expose on the highlights.

    One GH2 had a 200mm zoom using FM2.1 and the other, a 50mm Prime with Cluster 3. This was 720p and I was overcranking the shutter because I didn't have my ND.

    The Cluster 3 did lock the camera up once requiring a battery removal, but out of 60 gigs of videos, not too bad.

    When I started cutting the FM zoom in, in post, it had a noticeable grain compared to the Cluster. Thinking the lens may be the reason, I did some test clips with the Cluster camera today comparing the zoom to the prime and there is no noticeable difference in the grain (I didn't pixel peek).

    The differences in yesterdays shots seem to be when I over exposed a shot. Cluster seems to handle that better than FM or at least at higher shutter speeds up to 400.

    Any thoughts about that?

  • What's the differences between Sedna Q20 and AQ1..i read that they differ in quantisation but how does this translate in real life scenes? Better in low light e.t.c?

  • *** NEW *** Cluster v4 Improvements in 720p sustained hi bitrates with most importantly good buffer retainment. 70m+

    24p Always good at around 116M. Cluster in 24p H mode will span completely to roughly 1 hour 10 minutes on a freshly formatted 64Gb 95m/ps Sandisk card. Test confirmed.

    HBR 60M + sustained.

    Back to more classic looking charts!

    Cluster v4 - seti.zip
    1K
    Cluster v4 -78M constant average bitrate - 720p60.png
    1290 x 678 - 255K
    Cluster v4 -116M Average bitrate 24p.png
    1289 x 671 - 252K
  • @soulkeeper Yep, definitely resolves more detail in the shadows in low light on AQ1. It just pushes it a bit more from the Q20 to samples of around Q16 when required.

  • @Driftwood Cluster 4Looks pretty good so far. The detail looks good and I couldn't get it to lock up even using a 45mb card. I got about 114mb in 24p and 70 in 720p.

    I need to render it to see if there are any panning artifacts which there were not in v3.

  • @andyvia1968 You have a good musical taste, but do you mind putting the due credits as to who the musicians are and maybe a link to their itunes or something would be nice gesture!

    @Driftwood man, they are coming fast ; ) Which one of the clusters would you recommend for the low light/high iso?

  • GH2 with Cluster v3, Olympus 45mm 1.8, cinema filmmode, 1600 - 2500 iso

  • Is Pictoris 24L still the best all-intra/low bitrate setting that spans on lowly Class 10 cards? (i.e. Sandisc Ultra 64Gb with write speed 10? Mb/s)

  • I just got my GH2 the other day and hacked it with Orion v4b. Thank you so much Driftwood for the amazing work you do!

  • @oscillian At the moment its very close between SMBU v2b and β Pictoris. Remember there are four flavours of β Pictoris to choose from. Experiment.

  • @nchamberlain13 Thats a very nice start for a new GH2 adopter. Just remember to watch your ISO bugs as things can get a bit noisy on 320 / 640 ISO unless you switch down from a higher setting. And Welcome Aboard GH2 land! :-)

  • @driftwood thanks for explaining the diff. Q20 vs AQ1..

  • Wheather was sunny and more than a fresh breeze today (~6 bf), so I spent (probably too many) hours testing some patch settings.

    I first shot some reference material with GOP3ZILLA (which I have used a lot already), next I installed Span My Bitch Up v2b, and indeed this was the first GOP1 setting that was (unlike the other GOP1 settings I had tried before) stable enough for me to do some actual quality tests.

    All tests were done on a freshly formatted SanDisk 64GB 95M/s.

    Span My Bitch Up v2b Test

    1080p "24H" mode

    Recording the "moving tree leaves in the sun" scenario yielded a bit rate of ~100 MBit/s in 24H mode with Span My Bitch Up, the result looked very similar to the ~95MBit/s 24H recording I shot with GOP3ZILLA before, with a very slight favor that I would assign to GOP3ZILLA because of some minor "staircase" artifacts I saw in the "Span My Bitch Up" recording near thin vertical branch lines. I also tried 24H mode with ETC, and there were some very visible "horizontal bands of slightly different color" in the less bright areas of the image - I had not seen such while shooting 24H ETC with GOP3ZILLA. Since this effect is better visible "in motion", find attached an animated GIF for illustration - look at the tree trunk in the lower middle.

    The "24H" recording stopped involuntarily after 3:23s with the usual "card speed" message.

    720p60 mode

    SH

    Recording the "moving tree leaves in the sun" scenario yielded a bit rate of ~87 MBit/s in 720p60 "SH" mode. The IQ looked very good, but not visibly different from the ~50MBit/s 720p60 recording I had done with GOP3ZILLA before. The first 720p60 recording went fine, a second and third take stopped involuntarily each after ~3 seconds with the usual "card speed" message.

    H

    I tried 720p60 "H" mode to see whether that was more stable in the test scenario. It is, but this mode yielded a bitrate of only ~22MBit/s, and the image quality was consequentially very poor. (I guess the patch setting wasn't really designed for that mode?)

    HBR

    My attempt to record a 30 fps HBR video resulted in an almost immediate camera crash, had to take out the battery for revival.


    (I also did some low-light test shots with Span My Bitch Up v2b, they will be discussed in comparison to Flow Motion 2.02 in the FM2 thread.)

    span_band.gif
    1920 x 1080 - 8M
  • @karl Stick to stock - you seem to like long recordings. Its funny seeing you test a cut down Intra patch against everything else. Hope you had fun. :-)

  • I prematurely declared Cluster v4 stable. I did have 1 lock-up at about 3 seconds last night but subsequent 1 minute tests at 30 fps/HBR ISO 3200 went smoothly so I wrote it off as a fluke. I also tested 30 fps/HBR in full daylight ISO 160 with adjustable ND Filter on blowing leaves, pans and zooms and had no problem.

    Therefore I loaded it for a trip today at Foxwoods Casino (I kept V3f on the card just in case. To my amazement while shooting what looked like a non-stressful sequence (20 second pan across the floor and up a wall ending on a skylight dome) I experienced 3 failures (2 write errors and 1 lockup) out of 6 attempts (And I was quite careful with pan speed on the 3 that succeeded). Average total bit rate on the successful shots and the write error shots hovered around 62 mbps. Average I frame rate (once it got up to speed hovered around 450,000 bytes.

    To add to the mystery after I got home I tried 30 fps/HBR on low light and the out doors shots again an no errors with not much change in bit rate and I frame size....

  • I don't know Bonzai....I had a good night with Cluster 4. No lockups and very fine grained noise up to ISO4000.

    Today it did well until I left the polarizer on after I went in deep woods. IQ suffered badly but that's not the patch's fault.

    I was testing flow Motion 2.2 at the same time and surprisingly, it locked up once and required me to remove the battery. That was on a light colored tombstone and Cluster 4 handled it without problems.

    Spanning is low on my priority list (30 second clips are 30 second clips) so I didn't try that and I didn't have time to test it on running water which locks up everything (almost).

    So far it's been stable with good IQ. I do wonder if it's any improvement over Sedna though.

  • About to load up a 1 hour 10 minute Cluster v4 24p span on a bush for Karl...

    I dont give two fucks about spanning on Intra settings with top quality, if these guys want it, they can can stick with other people's settings - I dont give a damn.

  • A frame grab from last night...Iso 3200 using a 400mm telephoto.

    Photobucket

  • Zzzzz Guys, youre never gonna get a brilliant HBR or 720 setting on the GH2 - period. Its simply not powerful enough... Dontcha thiink we'd have found it by now? Its not bad but it will NEVER be brilliant. Take for example 720p60 - it goes shitty as soon as you go over a Max GOP Size of 1,000,000 Bytes.

    If you don't believe go and create some settings.

  • @Driftwood I'll take not bad but better than stock in 720 and brilliant in 24P!

  • @driftwood - amen. What you've created here is a game changer. And its not for boob tube or for U tube. Watch the stuff on a big screen! The tangles might persist between users and yourself, but the end result is this - THE FECKIN CAMERA IS MUCH BETTER thanks to Vitaly , You - Driftwood and all the other brilliant techno-rotters involved. So thanks! With a capital THANKS.

This topic is closed.
← All Discussions