Here are some screenshots of a short movie made for Tropfest I DP'd with GH2 (Dark Matter v3):
http://www.reddit.com/r/Filmmakers/comments/uvzep/48h_upload_notice_preview_screenshots_of_a_movie/
@rajamalik It will be great in less detailed scenarios, however why do you think I left Sedna at Q20 / AQ1? Because in higher detailed scenes the Quantisation scale will rise to greater variations in Q - this has been discussed many times before - the greater the difference in Q between one macroblock to another adjoined will show up artefacts.
@kazuo Probably not good enough card by the sound of it. It should be fine on cards showing Extreme or greater than 30M/ps
@thepalalias any recent tests???
I was recently comparing the Cluster v4 detail level on 1080P/24 and 720P/60. Here are a few samples for you guys. Handheld, fixed focus, OIS on. Standard -1 -2 +1 -2. By the way the third sample is the same 720P nicely upscaled to 1080p. 24P detail is fantastic as you can see! But 60P motion was fantastic, too :)
Is Cluster v4 pretty reliable in 60p?
@thoughts2uk A lot of the Quantum settings also have very good 720p60. As in this old chestnut.
@driftwood I only checked shorter clips.
@driftwood Not true, FCPX plays the Cluster v3/v4 footage back just fine. Don't underestimate it.
@nomad, thanks. I sorted the problem. Just switch camera over to NTSC, and now the flicker is gone in both the 1/50 and 1/60 mode. Strange that. In PAL it was there in both the 1/50 and 1/60 so had to choose 1/40 but in NTSC both the 1/50 and 1/60 work no probs.
1/ Here's the link to Cluster v4 BUSH Span test at one hour 10 minutes as proof it DOES span, is NOT corrupt, and works! This is Long GOP hi bitrate - most players or NLEs are NOT used to this sort of bitrate. Premiere Pro is the only NLE which can handle it, Vegas may do too.
This is still undergoing HD mode conversion by YouTube. Watch the birds fly in at 14:10 mins in.
2/ HBR 30p macroblock/tear test... does it? NO.
@driftwood How can I tame Cluster v3? Does it work the same as Sedna (divide Frame Limit by 8 to get Frame Buffer 2?). I tamed Sedna to make it 100% reliable without card write errors on my Transcend 32GB cards and I want to do the same for Cluster v3, which is GOP 6 instead of GOP 1 so the same theory might not apply.
@feha I have been using Cake 2.3 as well for more than a month now. I feel like it's an under-rated setting. If you are looking for image improvements with perfect reliability in all modes then Cake 2.3 is the only setting I know of currently that fits the bill. I use a 32GB Sandisk 30MB/s card and I have never had any stopped recordings in any mode.
@thougts2uk take a look at Cake 2.3 , I like Cluster 4 and FlowMotion 2.02 they have better handling for noise and saturation but i found most reliable Cake 2.3. @driftwood s settings are aimed for people who strive for top quality, not always spanning ...
ok cool. I'm in Korea now and is 10pm. got crew call at 7am so anytime so will be looking at this post when i wake up.
Also I noticed something interesting I have a PAL camera and shooting interior in 720p 50fps in Korea (NTSC Land) I have to put my shutter speed to 1/40 to stop the flickering. 1/60 the flickering was still there. Is this normal as the lights should be 60Hz?
Should I change the option from PAL to NTSC in Ptools? Will that help?
Thanks
Thats possibly a deficit in the buffers on hi detail - Ive just uploaded a 1 hour 10 minute span on Cluster v4 using Premiere Pro as the import (into AVCHD 24p sequence) which diidnt display any of those problems.
However, viewing the original mts file span files in vlc, clipwrap and an import attempt into fcpx didn't work out or displayed some problems like Benibube's. I am of the opinion that certain proggies can handle and others can't. Yet, my import/export from PP was perfect.
Playing the 'original mts' file back on some video players proved these proggies couldnt handle the GOP/bitrate - this is constant 116M remember!
Longer GOPs are known to be quite taxing because of the processing time required to look at the b/p predictive fields/frames. Intra doesn't have that problem.
I'm just going to do a outside test now on severe / dense trees in sunlight to see what the problem is and try to reproduce Benibube's problems.
Hang in there and Ill be back soon with a report. Plus, Ill be looking at Sedna 720p versus Cluster's 720p.
Its important to remember, both Cluster v4 and FlowMo v2.02 will stop recording if things get too taxing. Both will still write an mts file upto the recording stop point. Cluster 720p writes higher bitrates.
True @HillTop1 That is the way to reignite the inner decoder - memory clear!
@driftwood, I was using cluster 4, 720p for a shoot. When I play in camera I get card error however, if I power off the camera and remove the card and insert it back again, those same 720p files that previously gave me a card read error played just find. (I'm using the 95mb 64gb extreme card)
@driftwood but what about the corrupt files above with cluster 4?
Is it possible for you to email me a Sedna A AQ1 and Cluster v4 720 together? So that way I don't have to keep re-loading the firmware between shots?
Is the Cluster v4 720 50p better than Sedna A AQ1 720 50p?
I will be shooting the rest of the video in 1080p Sedna A AQ1.
It looks like you're new here. If you want to get involved, click one of these buttons!