Driftwood Cluster X Series 4:│Moon T8│Spizz T7

gittzy

Both cards are SanDisk brand 64GB cards @marciocons, under similar stress tests the MICRO SD Extreme 64GB 45mb/s in its card adapter performed flawlessly under all the same conditions. The new 64GB Extreme PRO 95mbps card stopped recording in VBR <<<%80 modes, and HBR 1080i, and had audio rendering, playback, and import problems. These experiments surprised me, because I would have expected more stable performance out of the newer, faster full size SD card, when instead the Micro SD 64GB worked flawlessly in all the same tests. I did my best to force it to crash or stop. I'm testing spanning now on both cards in Moon T8, then I will test Spizz, as well as Classix and report back differences on the cards and performances. Hope I don't raise more questions than I can answer and that this ads a positive variable to stabilize these amazing new GOPs.

marciocons

@gittzy Is the card a Sandisk? What is the problem with the card? Is it original?

gittzy

Bug Test:

Standard of tests: ALL NTSC video modes (VBR, HBR, FH, FSH, 24p, 24H, etc) for at least 5 minutes continuously. Racking Aperture and lighting conditions, zoom and movement. Brand New 64gb 95mb/s Extreme Pro card purchased from B&H delivered yesterday. Year old Sandisk Extreme 64gb 45mb/s microSD card with San disk included adapter.

Results:

SanDisk Extreme® microSDHC™/microSDXC™ UHS-I Card 45mb/s MICROSD Moon T8:
No stops within 5 minutes test of each mode. In camera playback functional, clips loaded into premier, audio Working working fine.

ClassiX Intravenus v3 matrix: No stops within 5 minutes test of each mode. In camera playback functional, clips loaded into premier, audio working.

SanDisk Extreme Pro® SDXC™ UHS-I Memory Card 95mb/s SD Moon T8: not yet fully tested yet.

ClassiX Intravenus v3 matrix: VBR 80%: Film stopped instantly when introduced to complex subjects, high contrast lighting, zooming an and out rapidly, panning or dollying. HBR and 1080i modes: Intermittently stopped filming within 1-3 minutes twice in 5 attempts, couldn’t pin down what caused it. No in camera playback, No audio in VBR 80% clips when loaded on the computer.

next test: complete T8 tests on 95mb/s card to see if there are any additional errors.

Spanning test.

Basis of these findings: While waiting for the 95mb/s SD card that I ordered from B&H, I began testing T8 with my micro SD card and didn’t encounter any of the problems members have been addressing this week. The 95 card arrived yesterday and the Micro SD card behaved surprisingly better than the new 95mb/s card did (though I haven’t performed a spanning test in all modes on either yet, I want to narrow down what’s specifically problematic first.) Why would a 45mb/s MICRO SD in an adapter card record more reliably than a 95mb/s SD card?

Empirical observation: Micro SD card was not nearly as hot after filming tests

Additional Tests: will control test temperature with a a laser thermometer after identical footage and check differences.

Thanks everyone, Have FUN!

Richard_Blood

I'm using Sandisk Extreme Pro, 95MB/s 16gb. All modes work except 720 60fps SH and variable movie mode in 720 when using <<<80%. I get the "recording was cancelled due to the writing speed limitations of card". Which is odd because the 1080 files have much more avg bit rate and don't fail.

producer

@cjdincer: Oh, the colors are not only less, they are drastically changed after pressing the record button. That time the whole picture becomes blown out, blue and red almost disappear and only greyish/greenish remains, no matter what WB was set in advance. Also, about the out of focus backgrounds: they seem to be lost of details when focusing on them more than the previous patches I've tested. Again: I'm talking about indoor situations.

cjdincer

@producer yep! you're right, I guess your deep smoothness is what @driftwood says "a finer low frequency matrix that strives to knock out artefacting at the expense of QP and higher bitrates." finer looking shadows and out of focus backgrounds. Also colour change seems less than T7 when you hit record or is it me?

producer

I tested Moon T8 in different situations and what I notice in every test is a presence of a lot of deep smoothness. It's like a Neat Video pre-applied with high settings, even when GH2's NR is set to -2.

See the frames before and after (raw and graded) below. Anyone with similar observations?

DarkwinQuark

Driftwood, so great you're back!
I've tried the ClassiX hack and shot in a mostly dark environment (concert hall) with an auto-focus controlled lens (will try a manual focus lens in the next days).
For me, spanning worked in 1080p 24H (SanDisk 64 gig), but I got my camera to crash once when I was shooting 24H in Ex. Conv. Tele mode (I did a side-pan right before the camera crashed and almost exclusively had non-lit audience members in front of me). Can't play that file now. But I was shooting a lot in the Ex. Tele mode and that was the only time it would crash.
But the hack looks great! Haven't tried much grading. Will get into that later.

matt_gh2

@driftwood Timing on release of Moon T8 and ClassiX w IV3 is perfect - I've got project Im shooting in a few weeks, so I'm squaring away look now. If I edit IV2 to add scaling tables from ClassiX setting, will it span (or would test be necessary)? Thanks!

driftwood

Driftwood ClassiX - with Intravenus v3 matrix (4,4,4,127s)

80Mbs 24p (12 GOP classic) Constant with a 4,4,4,127 hi/low frequency matrix - nice 'fine' filmic grain in the dark! Uses a lot of processing power but keeps a constant 80Mbps (quantiser changes automatically to meet picture requirement)

50p/60p (6 GOP) 50-60Mbps Variable

HBR/1080i/25p/50i - 40Mbps PAL (12 GOP),

HBR/1080i30/60i - (3 GOP) 60Mbps avg NTSC

Very high quality matrix rendering artefacts out.

Notes:

Use Low modes for spanning.

24pH will give you 7mins10 secs avg recordings per 3.99Gb.

BEST EVER LONG GOP 24P? BEST EVER 720P / HBR MODES with hi quality fine grain matrices.

TRY IT AND REPORT BACK. THANKS

Intra Freaks: If you want to try out the 24p matrices found inside 'ClassiX' on the old All-I Intravenus v2 - open & edit Intravenus 2 and copy over the scaling tables from this setting into it. Enjoy.

rojg

First time poster..

Tried Spizz t7 with sandisk 95mb 64gig card on FSH, looks great, no spanning though.. more tests when time permits.

TATZU

@driftwood how can i keep the bit-rate from dropping as low as 70mbps or less on 24h moont8 i would like to keep it around 140+

T1000

@driftwood

Some questions about Moon T8 and Drewnet 10:

Aside from the bitrates what are the differences between Drewnet 10 and Moon T8 then? Does one grade better than the other? Would the Moon variation of Drewnet 10 look and grade better than Drewnet 10?

Thanks for all your continued work on these hacks!

brudney

@x_worpig_x Driftwood already said it would be around 280Mbps - that's crazy high ;)

brudney

@driftwood thank you and great news! any chance of 6gop nebula variation of IV3?

x_worpig_x

@driftwood If you made an Intra variation, would you still have a constant bit rate of 80Mbps? Or would it be much higher?

Shaveblog

@driftwood Are you seeing the same import error with FCPX and T8 I described or is it just me?

babak

@driftwoon thanks for your feedback and explanations. I noticed spizz T7 is the best for shooting 25p. But as I said it just doesn't span in my camera. (BTW this T10 picture is really breathtaking as you say.)

driftwood

@babak Intra HBR 25p setting = just not enough power in the GH2 because its progressive structure is wrapped in a interlaced picture and HBR logarithms takes around twice the processing economics out of the codec than 720p. HBR30 being 3 GOP isn't as bad.

For HBR recordings try Spizz T7. I believe its the best looking HBR (3 GOP with B frames) and 720p (6 GOP with B frames) around at the moment.

DREWNET T10

Im just finalising the new DREWnet T10 'ClassiX' GOP (12 PAL/15NTSC), the 24p setting is using a newly developed 'Intravenus v3' matrix (4,4,4,127,4,4,4,127,etc...) which renders all recordings at a constant 80Mbps (the quanitser automatically changes dependent on the detail. This matrix is the finest (non coarse) hi/low frequency matrix ever implemented on the GH2.

Its extrememly powerful and can only be created on the DREWnet setting to achieve bottom QP of 14 on a 12/15 GOP (An Intra variation would cost around 280Mbps to achieve this so its impossible to implement on a GH2 - however, a frame /bitrate limited moon setting for close up/mid shot/ less detail shots is possible - requests accepted!). Skintones are breathtaking and grading is a pleasure. It'll be released soon.

Here's a picture of a 24p DREWnet T10 ClassiX render in Elecard Streameye.

babak

Thank you @marciocons for your explanation about disabling 'pasadena'. I still have to try that. I too wonder if that will help the spaning problem.

babak

Thanks so much @driftwood for your great work here. Here are some cropped stills from short clips I shot with T8 and Spizz T7 (taken from transferred footage in FCP7 timeline – footage shot on sandisk 95mbs 64gb, with canon fd 50mm f1.8 lens, aperture on 5.6). I have been specially looking at HBR qualities. There is an HUGE DIFFERENCE between HBR and 24H on T8 which is very surprising. The HBR footage looks very pixelated around the edges and very noisy, practically unusable. These differences were less extreme with Moon T7. With Spizz T7 the HBR footage looks much better than T8(bit rates: 71.3 & 68.5mbps). But unfortunately Spizz T7 doesn't span...

AshRich

Long time stalker 1st time posting. Tested Moon T8 today with a Scandisk Extreme 45MB card and some FD lenses, recording doesn't span, playback worked fine on files below the spanning mark but not the files that fail to span. Looks amazing as always thanks to the genius that is @driftwood.

Doctor

@Shaveblog I agree with you except I enjoyed seeing the videos of what people created with Driftwood's work, even if they were post processed. It showed some of the potentialities of the hack regardless of processing or content.

Shaveblog

If you're seeing significantly lower bitrate with Moon T8, you may be shooting 24L rather than 24H. Even if you didn't change any settings on your GH2, for some reason this happened to me too - I shot some test clips of what I assumed was 24H after reinstalling T8, but when I checked the bitrate it hovered around 70-80MB, which is what you should see from 24L clips using this patch.

I normally have 24H assigned to the GH2's C1 setting on the mode dial, with C2 set for 24L and C3 set for the "16mm Tri-X B&W" Ex Tele trick. So I went into the camera menu and set everything all over again to my usual 24H settings, assigned that again to C1, and shot some new test clips. This time they were all around 130-150MB as they should be with T8 at 24H.

If you're seeing very low bitrates and you think you're shooting 24H, go ahead and reset your camera manually to 24H and see if that doesn't sort things.

TATZU

@driftwood thanks for clarifying! And thanks for t8! I notice i am getting pretty low bitrates with this hack does that mean the quality is less than t7? is there a way to make the bitrate equal or higher on average than t7 i have plenty of space on my sd cards and want the best quality possible out of my trusty old GH2 =D

Howdy, Stranger!

Categories

Tags in Topic

Top Posters