Personal View site logo
Driftwood - Experimental Series 2: Low Rider, Cluster v8, V9, Intravenus II, GH3onaGH2, AN, Boom
  • 1008 Replies sorted by
  • @defier Dynamic ad too much noise over Standard or Cinema.

  • @defier "is there something wrong with Dynamic?"

    That depends on your goal. Cinema actually has the most linear curve (now there's an oxymoron), all the other profiles add something to that. Adding color, raising gamma, bumping the lows, etc.

    Now, there's nothing inherently wrong about adding those things in the camera instead of in post. It can even make the job easier on occasion. So if it works for your material, go for it.

  • @driftwood. Thanks a lot for the explanation and the tips. They are very helpful. With what I have learned from personal experience and the knowledge I have gained from you and the others, I know which settings I want and how to use them now. Cheers.

  • Thank you for this very detailed description, Nick!

    To be really flexible and quick on set, we'll need more than one body with different settings, I'd say.

  • @Zaven13 Youre absolutely right for wide shots high detail. At 24p Intra Low Rider if you edit Q to default 20 (uncheck all the quantisation stuff) will generally give an overall picture quantisation range of Q18 to Q22 for the whole picture - which is still good. [Sedna Q20 is a good example of this method]

    As soon as you start lowering the Q, employing AQ, or bringing down the matrix numbers for low pass freq filtering I guarantee Q for the whole picture will begin to vary - and sometimes more wildly. For shots like in tight subject matter (e.g. faces) a much lower Q can be used (such as Low Rider variation 1,2,3 or 4). Intravenus is also a very good setting for hi quality mid to tight shots/less busy shots/panning movement.

    Generally, for 24p I use two of my settings when filming; e.g. Boom or Intravenus for mid-in tight shots and any of the same setting with Q set to default (Untick all Q related stuff) for high detail wide shots. That way I have the benefits of maximising my picture in close (at something like Q16 for Low Rider, Intravenus etc... which is of course better than Q20).

    The rule of thumb is if theres less going on, encode the picture with the best detail you can. If there's more going on try and keep the quanitsation constant across the whole picture - and simply setting and experimenting Intiial Q (IQ in ptools) will help you achieve that.

    There is no way of knowing how you're going to get the best quality with settings (and Intra!) unless you experiment. But in summary, its always best to share the bitrate out throughout the whole picture whilst you can and then measure with elecard.

    I've done enough testing of settings and with the various way of lowering and keeping constant Q to know that no single setting works for all scenarious. And that includes other people's settings. Quantisation generally works from the top down and as soon as it starts running out of bandwidth the lower part of the picture will begin to suffer first with higher or poorer QP values.

    With Long GOP the bitrate expectations are much lower (due to prediction techniques) and Cluster v6 and v7 (6 GOP and 12/15 GOP) are exceptional all rounders providing some of the best looking constant picture quantisation Ive seen.

    However, each method Intra or Long GOP have their benefits (and fans!) whether its movement (I prefer it in Intra) or editing in the NLE (Long GOP takes a lot more arithmatic to decode and can be time consuming where as Intra is very quick as each frame is a picture). The arguments for each have long been discussed on these pages.

    The other problem is 50p/60p modes and FSH 1080i60/50 which have less memory and cpu power overall for the extra frames. Its for this reason why picture frame sizes in bits are smaller, etc... and a lot more work has to be done to balance things, The GH2 is brilliant at 24p and okay for everything else.

    Therefore I continue to support both Intra and Long GOP methods. Thank you once again for your excellent testing.

    The same can be said of the GH3; in 'tight shots' utilising the Intra settings is very good (Ive measured them!); zoom out and go wide with high detail and you're better off using the 50Mbps settings. Only when we hack the GH3 will we know if the High Profile level 5's use of 8x8 Transform will help when we push the bitrate up whilst trying to maintain decent overall Q for hi detail wide shots (and act well as a single setting for all). Its good practice to organise your shots from your storyboard with the right setting.

  • Yes: it is too unpredictable!

  • @driftwood. Intravenus V2 is exceptionally stable (I've only been able to get it to fail once after shooting for a few minutes). Whereas CM failed quite often (1 out of 15 shots or so). Would it be ok to push it past 150mb? Say 170 or 180mb/s like some of the older hacks to get even more detail out of it?

    And I posted this is my other thread, but my tests with a 50" plasma are showing Dynamic mode to produce a richer/more quality image than both Standard and Smooth film modes. I posted earlier that standard vs standard CM looked better than IntraVenus v2, but shooting IntraVenus v2 in dynamic is about the same. Basically everyone recommends shooting Standard or Smooth but is there something wrong with Dynamic?

  • @driftwood. Nick, I think I am ready to throw in the towel for having an all around setting with Intra 24p and stable FSH/SH with very good image quality. For those who mostly use 24p and HBR, the Low Rider original version gives you an excellent quality for 24H at 140MB and HBR at 105MB all intra. But FSH/SH will not work. To fix the FSH/SH stability problem it is necessary to reduce bitrates and frame limits and by the time it is stable, image quality has been compromised. A good example of that is the amount of time and effort @bkmcwd is spending to stablize non-24p for Valkyrie settings. Also as @shian indicated in his test video which included GH3, the 70MB all intra footage of GH3 sucked while 50MB IPB looked great. Maybe Vitalie's hack of GH3 will change things for the better and give us more options but for the time being, anyone interested to have an all around setting for all modes, you are better of with longer GOP setting like Cluster series, Flowmotion, Valkyrie/GOLGOP13 or Sanity.

    Nick, I would also be interested in a CLuster V8 3GOP or 6GOP settings based on Low Rider with 3GOP as preferred option.

    Thank you for all your hard work you have done and still doing to help me and others out.

  • @driftwood: I'm also very interested in Cluster and generally in any long GOP settings you may publish.

  • @driftwood. I tested Low Rider variation 4 for NTSC 60fps. FSH/FH and H survive the indoor test which is pointing to the TV and record with some zooming involved. SH still fails. However, FSH/FH, SH/H all fail the outdoor tests which is the street with trees, cars, houses in sunlight. HBR is very solid and it even survives the death charts.

  • In your examples, the Intravenus v2 holds more detail.

  • Canis Majoris Day Pm http://i.imgur.com/KutK5.jpg

    vs Intravenus v2 http://i.imgur.com/ZXeY2.jpg

    thanks Driftwood

  • @Driftwood, as promised I tested Lowrider 2 and 3 today using a variety of cards. The tests were simple, involving a manual lens, and various settings while shaking the shit out of my tripod and moving it all over and also doing static shots of a deep shot of allot of people in motion. The cards I used were an 8gb 30MB/s sandisk card, a 16gb 95MB/s sandisk card, a 16gb Toshiba Exceria 90MB/s read/30MB/s write card, and finally a Silicon power 16gb class 10 which is about 20mb/s. The results were Low rider 3 passed all tests including ETC and 720p except on the Silicon power card where it failed repeatedly during every setting, even during most static shots. However this is not a top card, so I call this a pass for Low Rider 3. Low rider 2 however had troubles on all cards in 720p, and ETC, plus in 1080p if I raised the film speed above 1/60th it failed frequently. Not a big deal for many, but I use 1/100th shutter speed on 1080p fairly often.

    I hope this is useful, I am not a scientific tester at all. Oh, and I was in SH mode at all times.

  • i had a long Low Rider 24p test today...aaaand it was very very stable without probs...!!!

  • @Zaven13 Can you check PAL settings for FSH and / SH ?

    NTSC 60fps FSH/SH adjusted. Try this 'Low Rider variation no. 4' for usual problems.

    Low Rider V4 with FSH-SH 60fps adjusted - setd.zip
    1K
  • Thx, Apocalypse Now 'Intravenus' v2 With cbrandin 'Smooth Cinema' Matrix is super nice

  • @driftwood. I am in NTSC world. I used Panasonic 14-140mm, OIS with Sandisk 64GB SDXC 95MB and Low Rider variation 3. Let me tell you the image quality is absolutely beautiful in all modes. I shot two different scenes. Last night I pointed to the TV with some zooming to cover the entire screen and this morning I shot my street with trees and houses and cars in the sunlight.

    • 24p and HBR are stable even with full zooming. Bitrates varying based on detail.

    • FSH/SH, fail in less than 10 seconds. The files still write out to the card after the stoppage. I used both IA mode and VMM and ther results are the same.

    • FH/H also fail in less than 10 seconds and have the same bitrates as FSH/SH. I don't understand. Shouldn't FH/H use lower bitrates therefore be a safer alternative? I would be happy to have the 24p, HBR along with a stable FH/H until the higher modes FSH/SH are further tuned. Thanks for all your efforts.

  • Driftwood, I will be out and about testing Boom a bit tomorrow, I will bring the latest Low Rider along for testing too, and will use a variety of class 10 cards to see how it/they perform. Consider it a done deal. But just to be clear I really want to see the extremes of my lens if only for 10 seconds clips. So I literally mean it when I ask, what is the all time highest detail patch? My guess is Seaquake or Sedna, but I am just guessing.

  • @Zaven13 Variation 3 gets by on the death charts, what subject matter/zooming/lens/etc produced the recording stop? And did the files still write out despite the stoppage?

  • @driftwood. I tested all three variations of the low Rider. There seem to be a common pattern in all three settings. 24p, HBR and 720p H modes are stable and look great. The failures occur with FSH/FH and SH modes. I have noticed a progressive reduction in bitrate usage from LR1 to LR3. FSH/SH are improved in LR3 since they use lower bitrates in lower detail scenes. But once detail increases and bitrates reach 65-70MB they fail. One thing puzzles me though. By definition, the FSH should use higher bitrates than FH mode but when I check the video, both modes use identical bitrates as if there was no difference.

  • @driftwood

    I have had a huge interest in cinema smooth since it's announcement. although I am a huge fan of intravenus 1, and now a huge fan on intra 2, i'm just curious to see what CS is like on it's own.

  • @GravitateMediaGroup Intravenus variation no.2 contains the Cinema Smooth matrix of course, and I am pondering over making a Cluster version of the settings - if enough people want it like @brudney.

    Low Rider is sorta taking things back to the roots a little and looking at what makes a certain matrix work well with Intra in 1080 modes and with a low constant Q. The three variations of Low Rider should be tested by those who want to test. All three variations work very well in 24p utilising different methods of rate control. HBR and 720p modes are still being tuned. Remember they are initially test settings and thus I haven't issued a release candidate until a few more people testing are happy. At the moment variation no. 3 is the safest bet with the variation 2 and the earlier released variation no. 1 possibly freezing here and there on FSH 1080i50/60 and SH.

    My own results show varying quality and I will talk about this soon.

  • Shooting a hip hop video in a club tomorrow...lowrider seems strangley appropriate. -lol

  • So many questions guys. Another one from me;) Any chance of 6/12 GOP versions of Cinema Smooth? cheers!

  • @driftwood

    was the "cinema smooth" by itself idea scrapped and will only be a hybrid with intra v2, or is low rider the same as cinema smooth?

    also, what are the differences in the 2 new low rider versions over the original? I know it's like asking a parent which one of your children are your favorite, but do you think low rider is a better patch than intra 2 and in what ways...low light performance, color, bit rate, etc?