Personal View site logo
'Apocalypse Now' Experimental Series 1 Thread - BOOM, Intravenus - cbrandin/driftwood AN Soft/Cinema
  • 1089 Replies sorted by
  • @BigBadFun try mysteron, sednas and canis. Orion v4b is one of the best but it is not the only best available.

  • Super impressive screen grab from today's shoot on a short. AN Boom! @driftwood - This is ETC on a 50mm OM lens. Gorgeous. The grain is SO incredibly fine and the chroma noise is definitely lessened. I'm starting to feel like Boom! is what the 7D wanted to be in terms of roll off AND resolution.

    Again, @retrospective, as with the previous screen grab, the source file is 1080p.

    image

    Screen Shot 2012-09-15 at 10.58.21 PM.png
    2560 x 1440 - 4M
  • People are trying to get the most out of the GH2 and put a great deal of effort into it. I can understand why emotions flare and a sense of competition comes into play.

    Flowmotion seems to be a more generalized hack and Driftwood has stated that Boom is specialized. Comparisons between the two are valid, but it would make sense to put the two in a variety of sittuations including ones that Boom is designed for to show strengths and weaknesses.

    In the end we all want the best performance from our GH2s and perhaps that means patches for specific sittuations or maybe a general hack is more practical for some of us.

    What I'm trying to get at is that we are all really on the same side here and I hope we can all come together and sing Kumbaya in the desert.

  • @drfitwood Glad you made it out. My friend had some close calls in Egypt. Deserts can be deadly. Great work with all these AN settings. Thanks.

  • @Meierhans I give it an 8.5. Good beat, fluid motion but hard to dance to!

  • As far as the sour thread is concerned - these particular hacks just happen to be made by individuals who clearly don't see eye to eye. I'm not sure why anyone would think these comparisons are not worth investigating...if you want a better hack don't continue screaming that the world is flat (444) - either do some quantitative tests or let the geniuses battle it out to the top of the mountain.

    That being said...

    Is there a particular thread on here that does have all of the comparisons to each patch with descriptions to which is good for what scenario? I think that would be helpful, especially to clear up any confusion that novice users have when they come to a thread like this and see allegiances being created (which really, c'mon now guys...) I honestly don't know how conclusive any of it could be, but it could help people create a more complete frame of reference for their own uses of their camera. I've used a good amount of different hacks at this point and have noticed the differences in IQ as well as each hack's 'character', but it's harder to say WHICH i would use in certain cases. Maybe I haven't read as much as I think I have, or perhaps we just need to make it more digestible for people? I'm not hugely informed on all the technical stuff but I'd get something like that going, even if it were just compiling videos and pointing out the notable differences between each setting.

    WHEN would I use flow motion over sedna, or over Boom, or Valkyrie (so far my personal favorite)? As far as I can tell I've only seen threads that are based on a specific suite of patches, created by the OP, with tests particularly targeted at making that particular set of patches better in their own right (slightly tangential to existing patches based on GOP length and IQ). In that way, the quantitative comparisons LPowell are making here are not so ludicrous to any others I've seen, particularly because Boom is very much an experimental patch with clear limitations. But perhaps there just needs to be a different thread for those comparisons, so the OP doesn't feel so slighted on what their topic ought to be (which is their hack). Perhaps clearly defining it for those who are less knowledgeable will also clear up WHY users should use a certain patch and HOW they can take advantage of the sheer amount of patches that are available to them (creating better GH users).

  • Orion v4b - still the best.

  • @driftwood I walked in "the death valley" fully high with San PEdro ( a cactus similar to Peyote). So i know what you mean! jajajaj also i´m from Chile, very glad you are enjoying the desert... awsome pics, is that a new hack? looks beautiful, many details and realisiric colors... hoping for a soon release!

  • I spent about 4 hours today at an event. Most was shot on my AC 160 but I shot about 20 minutes with Boom and a Samyang 7.5. This lens isn't especially sharp but it sure is wide.

    bod1.png
    1920 x 1080 - 2M
    bodparser1.jpg
    1600 x 900 - 824K
  • Need help - Can i use 24L on Drewnet patch and still get the benefit of the softer detail?

    I understand it will have lower bitrates, which is what I am after butdo i still get the benefit of the patch at 24l? (On cinema 24p mode).

    Thanks in advance

  • @driftwood are you using Boom on that desert shot?

  • BTW: Has anyone else walked the Valley of Death in the Atacama Desert? It nearly got me today. Ive never been so exhausted in all of my life.

    Driftwood in the Valley of Death - Atacama - Chile.png
    1280 x 1048 - 1M
  • @thepalalias Agreed, on every point you make. It's kind of embarrassing how personally some people take comparisons of their patches with others. Comparisons can be very useful when done well. The only way we can hope to understand the benefits of all of these various patches, is through differential analysis with other patches.

    • If you want good spanning results with GOP1 setting buy a SanDisk 64GB 95MB/s card. That specific card (not another version of it) is the one that has deomnstarted high spanning success to a degree that no other tested card has done. This has held true for months.

    , I see nothing wrong with posting comparions to FM2 that just show the data without speculating much about the rest. Everyone that read the description for "BOOM!" knows exactly what it is and is not and the comparison here is not aimed towards the strengths indicated and everyone knows that.

    We also know that there are many other Driftwood GOP1 settings that can handle that type of scenario better. So what is the harm of posting that data?

    If people are going to critique the comparison, do so based on testing methods, NOT based on whether the test should be conducted at all.

    One of the reasons the site has become home to so much improvement is because of the sheer amount of testing and comparison that people are doing. Let us assist each other in doing so in the best way possible, not discourage it from happening.

    Oh and for anyone asking: Sedna remains one of the most throughly balanced, extensively tested and high image quality settings. It is a good one to default to for high priority 24H shoots if you have the right memory card.

    Other settings can do better under some conditions. It all depends. Keep in mind that it mentions potential issues or things to watch out for in the descriptions for several settings: do not assume that those will be reliable in all situatuions, just that they will provide a desirable image in the situatuions where they do perform well.

    And remember, this is not a religion. I use Driftwood settings more than those by any other other. That does not mean that I do not use FM2 or Sanity, or GOlGOp, etc.

  • @driftwood

    Perhaps you should take a look at your setting against a comparitive equal - bkmwcd's excellent GOP3 suite.

    The notion that patches can only legitimately be compared to other patches that use the same GOP-length is hard to justify from an audience's perspective. When you watch a video, its GOP-length is irrelevant. It's the visual motion picture quality that counts. Regardless of the encoding technique used to capture a video stream, its image quality can be objectively compared to any other type of encoding.

  • Lee You think Ive ever required your assistance? Ive just demonstrated to you above that it maintains Q16 on average for the right kind of scene. Together with All Intra and exceptional grading qualities.

    I'm going to spell it out once again. ITS EXPERIMENTAL and performs better than anything Ive seen under the right conditions - so its a setting that I thought our p-v users could take a look at. Geddit? No one is forced to use it. In my quest for excellence Ive consistently found that fortune favours the brave.

    Perhaps you should take a look at your setting against a comparitive equal - bkmwcd's excellent GOP3 suite. Its my personal view after streamparser/streameye testing that theyre consistently better than FM2 if you and other users need to know - and resulting in finer Quantisation. I'm sure he could be of assistance to help you edit your future ptools patch settings.

  • @driftwood

    Your conclusion is bollocks.

    Granted that the Apocalypse Now Boom Flat 4s is experimental. I'm reporting the results of my comparison tests, which may help assist you in improving it. The point of using Flow Motion v2.02 as a reference is to demonstrate that the test scene is not unusually challenging and can be encoded with very high image quality.

    My observation of AN Boom Flat 4s is that it is demonstrably unable to maintain its target QP value of 18 over the full area of each encoded frame. As the Stream Parser report above shows, worst-case QP can range as high as 34, which indicates a significantly coarser level of quantization than the intended QP of 18. The unedited 200% crop screen shots I uploaded show an example of where this inconsistent macroblock quantization can become noticeable. Hopefully, these test results will provide insight on how the patch can be improved to maintain a more consistent range of image quality.

  • @LPowell Boom! was NEVER intended to be compared to FM2 or Cluster or anything else. Geddit yet?! Most people understand that. Comprend?

  • @flaschus @driftwood

    thank u for a comparison of apples and oranges yet again...

    There are no apples and oranges in the test footage, it's the same scene shot back-to-back with the same lens and camera settings in each case. If you don't think that's an "honest comparison" then I'd have to conclude that you're simply not interested in considering objective evaluations of these patches.

  • @LPowell Your conclusion is bollocks. Once again you compare only to justify yourself.

    Read the text on page one - its experimental setting for one, Two I suggested that this would be good for mid to tight shots especially facials. If you read this thread you'll also note that I mentioned if the data rate is lower than 145mbps you will get very good results.

    @all Boom! was designed to see how far we could push the Gh2 and quantisation.

    Here's an example of Boom Intra in action over here in Chile. This whole piece of footage never went over Q20 - indeed most of the individual macroblocks measured an average of Q16. Boom! is spectacular if you stick to my recommendations. But its not for all.

    @LPowell A few months back I compared FM2 to Cluster v6 and 5. I didnt see FM2 result in finer quantisation. Indeed Cluster v6 and 5 provided me with better Q even in my own examples of your so called flowing water test. Could I be bothered to enlighten you by commenting in your thread with this spectacular comparison? Of course not. I couldn't be bothered.

    Boom Intra mid shot test - never goes above Q20.png
    1277 x 657 - 439K
  • @LPowell,,, thank u for a comparison of apples and oranges yet again, a test that you say, can provide absolutes, about a 1gop (INTRA) compared to a 3gop patch (NON-INTRA)?!

    first, I say: ONLY SITH LORDS deal in absolutes;

    and, who cares about a Q number if it looks better in playback?

    btw, i did a REAL comparison of 3gop patches: flo2.02 and bkmcwd's 'soft 444'..(sep4. page 10 of this thread)....

    So, Y can't you make a 1gop patch and compare... or do honest comparisons of equal things.. thanks!

  • For those asking, "what's the point?" I'm giving you full access to the results of a straightforward comparison test of these two patches. It's not a scene that attempts to break either patch, nor was it selected to favor one patch over another. I've made the original video files available for download and independent verification of the authenticity of the test results. You can evaluate these examples on a completely subjective level if you wish, or you can delve as deeply as you like into the objective technical details of the encodings.

    For any comparison test you need a reference, and I chose Flow Motion v2.02 as a reference point that I understand in complete detail. In my view, FM2 serves as a good reference because it records this scene at its finest image quality, exactly as it was designed and intended to work. There's nothing uniquely suited to FM2 here, as this is not an unusually challenging scene and I would expect any well-designed patch to encode it without straining its resources.

    Frankly, I was surprised to find that AN Boom F4s was unable to encode this scene at its intended quality level of 18. The Stream Parser report above reveals that quality levels range as high as 34 in many of the macroblocks, which is a very coarse and inconsistent range of image quality. What that indicates is that AN Boom's Flat 4s quantizer matrix is demanding far too much bitrate to operate consistently at a QP of 18. Since the patch's 150Mbps is close to the peak bitrate the GH2 can reliably sustain, my conclusion is that either the target QP of the patch needs to be raised significantly higher (i.e. coarser) than 18, or the Flat 4s matrix needs to be replaced with something more realistic.

  • Valkyrie looks really nice. Very good setting. It uses a lot of bandwith though and did not span on my 45MBS sandisk ultra. It played fine in the camera though and i had no freezes.

    Its really tough to find the perfect run and gun, but still cinematic setting. So far flowmotion, cluster v6 nebular and valkyrie pure cbrandin 444 are my favs settings.

  • Nice comparison Lee, that confirms that Boom suits better for my kind of work.

    And that looks good Lambo, Valkyrie also deserves a try.

This topic is closed.
← All Discussions