Personal View site logo
GF2 Stable Settings
  • 614 Replies sorted by
  • I had a Transcend supposedly Class 10 and it seems not even to be up to decent Class 6 level. Very disappointing. I am having to upgrade all my cards to use the PVDOG patch even.

  • I also tried SpanMyB**chUp... but it didn't span.. but I'm not on confirmed working class 10 cards, liek the 30mbps sandisk. I'm on Transcend class 10..

    for me spanning is a must have

  • I just put pvdog back on and tested a 15+ minute record in 720p, worked great .. didn't try playback in cam.. but record worked great... used pvdog setf .. Im on NTSC.. dont know about PAL this patch is pretty dang good IMHO.. especially since it can work on old slower cards. Only 1 or two scenes I have thrown at it that I've seen any blocking and they were extremely motion and detail entrenched scenes..

    Now I'm trying to find the best picture settings for footage, been shooting very flat .. -1 on everything.. but don't think i need to go that drastically flat.

  • Is PVDOG now smooth and stable on 720p? (I understand it does not play back in camera).

  • Hopefully soon ClusterV2 and manual settings

  • Mysteron .. no spanning.. no good for me.. on to something else.. Footage looked nice.. if you can deal with every clip under 6 minutes... but a good test to see how fast the GF2 can write data.. 85mbps.. not bad

  • Well I still have Mysteron on my GF2, decided to record a test clip with a cheap $15 Transcend class 10 16gb card.. Recorded successfully.. clip is 1:23 in length.. 851mb in size.. 85,853 kbps footage looks nice .. this patch may eat up to much space.. at almost a GB a in 1.5 minutes.. im going to experiment and if i dont like it ill go back to pvdog. going to go test spanning now..

  • AndyS, yes Vegas does but the files would be way to big for internet upload with limit like 2gb on youtube 1gb on facebook etc.

    However to put this blocking issue to bed.. the attached image is a UNCOMPRESSED render still from vegas.. this is as good as the original, blocking is present but very very minor in the sky.. Vegas isn't destroying anything and when rendered to XDCAM HD the XDCAM codec went totally when bonkers on the blocking, and made it far worse. That made me paranoid and seeing blocks in everything I looked at .. yes there is blocking in the original footage in this extream shot, but its very minor.

    Uncompressed blocking minimal but still present.jpg
    1920 x 1200 - 1M
  • Ebach - Is it possible on Vegas to simply edit "native" the AVCHD and output edited clips in the original "native" AVCHD uncompressed? In other words, simply edit and output in the original Codec off the camera? Maybe "smart rendering" or whatever?

  • Agreed Pdlumina, one small change I made was it wasn't anamorphic, but 1:1 original frame size Either way yes mpeg4 is going to be more efficient for quality vs space used for any same file size.. But still shocked how badly xdcam trashed it in that area that was blocking.. Otherwise detail was good in the xdcam formated render.

    But enoght with this render format talk .. Time to keep testing and shooting.

  • @Ebacherville If what I read about XDCAM is true (35Mbps VBR, 1440x1080 MPEG-2), then something to remember is that (I am talking delivery here) MPEG-4 is a much more efficient and optimized codec than MPEG-2.

    With this in mind, I am certain there is a world of difference between 35Mbps MPEG-2 1440x1080 (anamorphic) and 35Mbps MPEG-4 1920x1080 raster. Certainly it would be best for the video to have the highest visual quality before being sent to be re-encoded by an online video delivery service.

    Meanwhile I still argue with the production department that raster 1280x720 progressive has more actual quality than the 1280x1080 interlaced (DVCProHD) perceptual quality.

  • @pdlumina, yes agreed, my shock was how badly the render to XDCAM added to the blocking.. mostly shocked because XDCAM is widely used and fairly trusted format.. As I said, doing uncompressed renders from the timeline didn't change the source image, blocking is very minimal in the source and uncompressed, but the XDCAM render made the blocking far worse. I think I'll use XDCAM renders to find flaws now, instead of render for the web. :) I messed with some MainConcept CBR settings and think I found a winner for high bitrate yet compressed renders.

  • @Ebacherville It is my understanding that these patches allow us the users to pump a higher bitrate into a 'delivery' codec when acquiring images.

    Because of this, I tend to either A) edit the original Inter-frame with an application that supports the original files (Vegas, etc), or B) transcode to a proper intra-frame editing format for a smooth editing experience.

    I tend to go with option 'B', and never go below 100Mbps on transcoding (DNxHD on Vegas or MPEG-2 I-Frame on Lightworks).

    It may seems like an overkill, but that's just the surface. I am going from a highly compressed inter-frame stream to a lesser compressed intra-frame format, and a simple understanding of the main difference between these formats (GOP's VS independent frames) yields the logical conclusion that I need much more bitrate to avoid losing too much quality on transcoding.

    On delivery (web, local proxy) I then go back to H.264. I tend to use 17-20Mbps for 720p and about 25Mbps for 1080i (rare). AAC for audio and MP4 as a container. This works wonders.

    Of course, sending the file to a third party online service (the tubes, vimeo, etc) means that you lose control over the final result, since these services batch-encode the users videos at a preset bitrate regardless of content.

  • At this point I don't think its a Vegas issue, because rendered from the Vegas timeline uncompressed it didn't change the image. I'm going to mess with it more and see what I can find out. At home I have a older version of Vegas, at work I have the newest version.. I'll do come comparisons between the two. Still shocked that XDCAM at 35mbps (highest setting possible in Vegas) accentuated it that much.. but I guess we are going from 50mbps to 35mbps.. some generation loss is expected.

  • Ebach - what version of Vegas are you running? If it is the new version, it is pretty shocking that it worsens these problems.

  • I have looked at this clip every way possible, it still has blocking in all, but less than the XDCAM format.. it is in the source clip.. still will continue with some other patches.. pvdogs is very good.. remember were talking about a $250 camera body here.. even uncompressed you still see it, maybe its just me now that I've seen it, and i know exactly where to look for it.. posted the YT file with a challenge to find this downfall.. no one has spotted it yet. May fool the masses. I still love this patch great video can be had from it.

  • My gosh I'm totally dumb founded.. here is a render to QT uncompressed of that 10 second shot from my Vegas Pro timeline.. blocking is totally gone.. Sony HAS indeed screwed me again.. this 10 second clip in QT uncompressed is over 2.5gb.. WOW pvdogs patch is going back on my GF2.. And I'll figure out a better way to encode video for showing to everyone than Sony's own xdcam @ 35mbps... it is absolutely scary that sony's own format is totally messed up! I think I'll bring this to work tomorrow and drop try it in final cut at 2560 x 1440.. should total prove it as a solid patch. my god totally unreal!

  • Wow, your correct I dropped the source file into a diffident program and the blocking is far diminished.. still slight.. but far better, attached is a still from raw footage.. I'm going to investigate this further.. could totally be my error. I think Vegas Pro has totally let me down once again..

    Copy of pvdog still raw.jpg
    1920 x 1080 - 424K
  • @Ebacherville

    These are stream frames from Mysteron posted by driftwood http://www.personal-view.com/talks/uploads/FileUpload/ef/2408936b64fc25841039e5db70377d.png

    These are stream frames from pvdgo's settings posted by me http://www.personal-view.com/talks/uploads/FileUpload/1f/753041f8833cf41fbe59315eed36ba.png

    Notice the 'health' of P & B frames compared to Mysteron.

  • Just loaded Driftwood's Mysteron on my GF2.. did a test clip.. no record errors.. did a test shot, averages 54,000,000 in low light with a 50mm 1.4, looks great.. This is my next test subject for hacks on the GF2 , and a GF3 once it arrives. I was using a fairly cheap class 10 16gb Transcend card.. further testing tomorrow.

  • @pvdog, finally stressed your patch to breaking point.. well it didn't break, as in stop recording. But was able to throw enough at it to make it do some major blocking .. the high motion shots of the water (like the first shot in this video, look at the sky , major blocking .. however only on those shots with massive motion detail.. the other shots that have sky are not blocking .. now that being said, considering I tested this on older class 6 cards and one was a class 4 I think.. and it worked perfectly, its still a great compromise for quality and reliability even on slow cards. In my quest for perfection of the GF2, I think I might go to a higher bitrate patch and higher class cards .. Might try cake 2.0 as feha posted above.. but still like your patch allot.. great performance even on slow cards unless you kill it with extreme motion shots that have chunks of non motion in them. Youtubes reencode lessens it , in my original footage, and 35mbps YT upload, its very apparent.

  • @GF2/3 owners!

    Since I don't own any of the GF line cameras, I'm recruiting testers who are interested in help me test 720p Timebuster settings: http://www.personal-view.com/talks/discussion/comment/51292#Comment_51292

    Thanks for your time.

  • I think that until Vitaity, does not come with the option of manual video, try all these settings, it is a bit useless ..

  • Here is my latest test GF2, Cake 2.0 settings, night shot ...

  • @Ebacherville Good footage. Manual setting is needed after all