Announcement

Collapse
No announcement yet.

How interested are you in a GH6 anymore?

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    Originally posted by Thomas Smet View Post
    You also want a Keyer capable of preserving detail like that for shadows and transparent objects like glass or fabric. Things like motion blur, glass and fabric can really get messed up by fudging with the values too much. Even if you shot with a certain shutter speed and added fake motion blur you still run the risk of crushing shadows and transparent objects. Thats why its better to get into the practice of keeping edges and details as natural as possible. If its a talking head static kind of thing like an interview maybe not a big deal but that really limits the creativity and use of a green screen to only use it for that kind of material. The real fun happens when you do full body virtual sets or have to move the camera.
    Hello Thomas,

    Yes, for most of my typical corporate or documentary green screen stuff it's not an issue - and for someone sitting or standing and talking at 30fps a shutter speed of 1/60 or 1/80 produces great results with Delta Keyer in Resolve/Fusion. But in cases like that there's very little motion blur to deal with - maybe some hand movement or a bit of wispy hair - and this can be dealt with fairly easily. Lately I've been playing with adding a Glow node (set very low) after the Merge -- to smooth older skin tones/details a bit and take the curse off the 4k sharpness - since on a lot of shoots since Covid the subjects (or company policies) have eliminated full makeup.

    But one specific shoot comes to mind where the motion blur without a faster shutter was crazy -- a full height shot with two cheerleaders doing 'jump splits' (I don't know what you call the move??) with pom-poms and their long hair moving all over the place. But you get the idea... The 30p (1/60th) blurs from a 'normal' camera were a nightmare -- but the 120p footage from the slow motion camera (w/180 degree shutter) proved a better option for us, even when brought to normal speed by frame skipping.

    Comment


      Originally posted by Zim View Post

      Remember when people would say, "Pros don't use AF" It still looks like a great camera and improves on the GH5 which was a great camera.
      I also remember when I had an operator for (almost) every camera -- at least every camera except for the wide establishing.
      But in the last year the number of shoots where I've been DP'ing and operating 3 or even 4 cameras to limit the number of people (due to Covid protocols) has made me wish for better AF from my GH5s and BGH1, and really see the need for an EVA2 with great continuous AF (to match that of the FX6 and FX9).

      Comment


        Originally posted by Zim View Post

        Maybe it will be a brand new amazing MFT line. I hope so. Until then I guess I use my 12 Pro Max and GH3! to make my next film. If the GH5II would drop in price maybe. I'm sure the AF is improved. Remember when people would say, "Pros don't use AF" It still looks like a great camera and improves on the GH5 which was a great camera.
        The GH5II is nickels and dimes. You should just get it and make your next film and then when it's done I'll go in and zoom in 300% on all of the parts where focus is dancing around. lol

        And people say a lot of things...it's why we've had world wars.

        Comment


          Originally posted by OnSet View Post

          Hello Thomas,

          Yes, for most of my typical corporate or documentary green screen stuff it's not an issue - and for someone sitting or standing and talking at 30fps a shutter speed of 1/60 or 1/80 produces great results with Delta Keyer in Resolve/Fusion. But in cases like that there's very little motion blur to deal with - maybe some hand movement or a bit of wispy hair - and this can be dealt with fairly easily. Lately I've been playing with adding a Glow node (set very low) after the Merge -- to smooth older skin tones/details a bit and take the curse off the 4k sharpness - since on a lot of shoots since Covid the subjects (or company policies) have eliminated full makeup.

          But one specific shoot comes to mind where the motion blur without a faster shutter was crazy -- a full height shot with two cheerleaders doing 'jump splits' (I don't know what you call the move??) with pom-poms and their long hair moving all over the place. But you get the idea... The 30p (1/60th) blurs from a 'normal' camera were a nightmare -- but the 120p footage from the slow motion camera (w/180 degree shutter) proved a better option for us, even when brought to normal speed by frame skipping.
          But that Motion blur shouldn’t be a problem. The Delta keyed on its own is not enough. You need a clean plate to even out the screen and lighting. Then the Delta keyer range doesn’t need to be as high and you can keep the motion blur.

          less motion blur is only easier to key with the basic view of using the range slider to get a good result. In that case yes less motion blur will be easier but that’s not the most natural way to do it. It leads to a lot of post processing that can make the shots look fake. Plus it’s much harder to work with and requires a lot more light. Adding motion blur in post takes a lot more processing power as well. It’s a solution but it’s not the most optimal solution and it is not the only solution.

          next time you have a shoot like the cheerleaders shoot a clean plate with the camera in it’s locked down position or just grab a second or so of the video when they move away from the green screen. You can loop a small segment. It’s all the same thing. Then Delta can use that clean plate to even out the screen and lighting. This gives you a perfect color screen behind your subject and when you apply the key color you can use a much smaller range that will maintain the motion blur. Fusion also has a clean plate node to help fake a clean plate if you don’t have any portion of just the naked green screen. It’s much better to have a naked section of video however. You do want to make sure it’s a small video section as well and not just a still frame. A still frame doesn’t have a moving noise pattern and may not look as good as a small video segment that does have noise that moves around each frame. Of course you should shoot as clean as possible to not have my noise at all. Noise can create edge chattering which means you have to choke or blur the edges even more.

          absolutle best key I ever pulled with affordable cameras was shot a raw 3:1 4K on a P4k and keyed with Delta and a clean plate. Every pixel and strand of hair was maintained even with fast motion and motion blur. The fluid movement of hands moving very fast looked exactly like it should have on a real background thanks to maintaining the perfect motion blur. It’s insane the level of VFX ability we have now with a $1,300 camera including the full version of Resolve/Fusion. If they could just make a global shutter camera at this price point it would be the ultimate VFX camera.

          Comment


            Originally posted by aram View Post
            Thanks for the ahalpert and for the clarification, Thomas Smet , I did not realise that it was about playing at 24fps but without slowing the footage down. That makes total sense!
            right, so what we're saying is totally irrelevant if you're definitely going to use the 60fps as slow mo. But I shoot a lot of stuff that is all 60fps just in case the editor wants to slow mo something. And two shots from the same clip could be used separately as normal speed and as slow motion right next to the other in the edit. So, I'm keen to experiment more with 120fps for that scenario. If I shoot is 1/120 shutter then I'm probably at the same shutter speed I was going to use for 60fps anyway
            www.VideoAbe.com

            "If you’re really in favor of free speech, then you’re in favor of freedom of speech for precisely the views you despise. Otherwise, you’re not in favor of free speech." - Noam Chomsky

            Comment


              Originally posted by OnSet View Post

              I also remember when I had an operator for (almost) every camera -- at least every camera except for the wide establishing.
              But in the last year the number of shoots where I've been DP'ing and operating 3 or even 4 cameras to limit the number of people (due to Covid protocols) has made me wish for better AF from my GH5s and BGH1, and really see the need for an EVA2 with great continuous AF (to match that of the FX6 and FX9).
              Absolutely. Obviously, you still need to worry about panning the cameras to reframe. But as far as keeping the subject in focus on unmanned cameras, AF is an absolute boon. And with wifi smartphone control, I can monitor my cameras remotely without any additional wireless transmission systems (albeit with some drawbacks (yet some advantages) over an additional system). I can rack focus to a different subject and have the camera track them using the smartphone app, for example
              www.VideoAbe.com

              "If you’re really in favor of free speech, then you’re in favor of freedom of speech for precisely the views you despise. Otherwise, you’re not in favor of free speech." - Noam Chomsky

              Comment


                Originally posted by ahalpert View Post

                right, so what we're saying is totally irrelevant if you're definitely going to use the 60fps as slow mo. But I shoot a lot of stuff that is all 60fps just in case the editor wants to slow mo something. And two shots from the same clip could be used separately as normal speed and as slow motion right next to the other in the edit. So, I'm keen to experiment more with 120fps for that scenario. If I shoot is 1/120 shutter then I'm probably at the same shutter speed I was going to use for 60fps anyway
                The shutter speed is where it gets interesting and where you run into a The Hobbit situation. Sticking 120p in a 24p sequence sounds logical until you factor in technically 120p would be shot at 1/250th shutter. That means each of those 24p frames would like they were also shot at 1/250th which isn't always optimal for normal motion and can look a bit jittery. If you want it to be slow motion you really do want it at 1/250th so each frame slowed down to 24p looks exactly like it would as if the subject was moving that slow in real life. But for a conversion its problematic.

                Thats why its really best to shoot slow motion as slow motion and plan for it if at all possible. I guess this is one area where faked motion blur in post could help a bit more. I'm just not a fan of it personally and think it looks fake. Many of us are trying to move more towards natural looking video and away from over processed.

                The same is true if you can't shoot higher framreate but still want to use optical flow in post for slow motion. If one can only shoot 24p on their camera and they really want a 5x slow motion they should shoot that 24p at 1/250th. That way the optical flow processing has sharper frames to use to create new frames and the motion blur will look as it should as if the camera could shoot 120p. Optical flow is based on motion vectors and predicting motion. The more motion blur there is the harder it is to predict and create edges.

                With that said we can't always know when and where slow motion will be used in an edit so we have to make do as best we can. I do try to plan for slow motion as much as I can so I can do it the optimal way.

                Comment


                  Hello Thomas,
                  Thanks for the suggestions.
                  I'll have to give it a try with some testing before the next music video with that director (who also has a tendency to put a lot of his performers in reflective silver and gold outfits -- always makes for an interesting shoot!)

                  Comment


                    Originally posted by OnSet View Post
                    Hello Thomas,
                    Thanks for the suggestions.
                    I'll have to give it a try with some testing before the next music video with that director (who also has a tendency to put a lot of his performers in reflective silver and gold outfits -- always makes for an interesting shoot!)
                    What codec are you shooting in? I feel like keying is the one time I'd really rather have shot RAW than 4:2:2. (I mean, I'd prefer it all the time. But it's the only circumstance I really regret
                    www.VideoAbe.com

                    "If you’re really in favor of free speech, then you’re in favor of freedom of speech for precisely the views you despise. Otherwise, you’re not in favor of free speech." - Noam Chomsky

                    Comment


                      Originally posted by Thomas Smet View Post
                      .

                      With that said we can't always know when and where slow motion will be used in an edit so we have to make do as best we can. I do try to plan for slow motion as much as I can so I can do it the optimal way.
                      This is true. For the music video I'm referencing, I had started out shooting all lip synching run-throughs as 24fps and everything else as either 120 or 60. But the performer was very inconsistent, and she also did some dance moves in-between lyrics that I thought could be useful in slow mo. I really don't like the look of footage slowed down in post, so i decided that 24p with a higher shutter was a lesser evil. (Can't remember if I used 1/120 or 1/250)

                      i consulted on the edit. The 2 most common slow motion speeds the editor used were 40% and 60%, probably used in equal measure. She probably used 20% and 80% in equal measure, albeit less often. There was one shot where she dropped below 20% (to 15% perhaps). I also feel like having more frames to work with probably yields better speed ramps in post
                      www.VideoAbe.com

                      "If you’re really in favor of free speech, then you’re in favor of freedom of speech for precisely the views you despise. Otherwise, you’re not in favor of free speech." - Noam Chomsky

                      Comment


                        Originally posted by ahalpert View Post

                        This is true. For the music video I'm referencing, I had started out shooting all lip synching run-throughs as 24fps and everything else as either 120 or 60. But the performer was very inconsistent, and she also did some dance moves in-between lyrics that I thought could be useful in slow mo. I really don't like the look of footage slowed down in post, so i decided that 24p with a higher shutter was a lesser evil. (Can't remember if I used 1/120 or 1/250)

                        i consulted on the edit. The 2 most common slow motion speeds the editor used were 40% and 60%, probably used in equal measure. She probably used 20% and 80% in equal measure, albeit less often. There was one shot where she dropped below 20% (to 15% perhaps). I also feel like having more frames to work with probably yields better speed ramps in post
                        It definitely does. 60p works much better for optical flow than 24p does for example. Less gaps between frames and less motion blur. Its much harder to have completely new random motion between each frame with 60p. 24p has a ton of gaps and something as simple as an eye blink can be eye open one frame and the next frame completely closed. Its impossible to have motion vectors for that because its like an on/off switch with no in-betweens to factor in. 60p allows that eye blink to at least have some movement of that eyelid. 120p will be even better.

                        Eventually we will hit a point where having higher frame rates for slow motion may be kind of pointless. For 24p projects 60p is already 2.5x slow motion which is actually a lot really. Especially for normal motion like people walking or day to day motion like getting into a car. Sports can need a bit more since the subjects move very fast. 120p is 5x which is typically enough for that. 5x starts to get really slow and can sometimes be a bad thing for our low attention span generation. Even a very basic task at 5x slow motion is a very long shot to watch. You have to have a very specific purpose for slow motion at that speed. Like a basketball slam dunk for example. Something that would be over in one second and now can be five seconds to better show the emotion of the act. A guy walking his dog at 5x slow motion is kind of ridiculous really and kind of becoming a cliche like lens flares and overuse of rack focusing were at one point. We are throwing things into slow motion now just because we can without any real aesthetic value.

                        Beyond that for like 240p really is silly and only really useful for engineers and science to study things. It has very little creative value outside of explosions which I hope most of us are not actually doing on a low budget. Because 120p is already very highly temporally detailed its pretty easy to optical flow that to 240p anyway. Unless your subject in insanely fast like a fiery explosion or bullet through an apple.

                        Comment


                          Originally posted by NorBro View Post

                          The GH5II is nickels and dimes. You should just get it and make your next film and then when it's done I'll go in and zoom in 300% on all of the parts where focus is dancing around. lol

                          And people say a lot of things...it's why we've had world wars.
                          As long as people can't see it at 200%!!
                          Okay I'll make films and you can enlarge other peoples work to 300%. Keep up the good work!

                          Comment


                            Originally posted by ahalpert View Post

                            What codec are you shooting in? I feel like keying is the one time I'd really rather have shot RAW than 4:2:2. (I mean, I'd prefer it all the time. But it's the only circumstance I really regret
                            Yes, this could have been part of the issue as well, since I was doing VFR 120fps with the EVA1 - which as I recall meant that for full sensor shooting, the sensor would have been set to s35/2.8k recording 1080 4:2:0. I do recall setting up the various camera settings on an sd card so I could quickly switch between 4k 4:2:2 for normal speed, 4k h.265 at vfr 60p, and the 1080 vfr 120p 4:2:0.
                            Of course I'd like to see more options and internal setup files for the EVA2 as well if it ever happens...

                            However -- for normal green screen shoots my corporate clients have been extremely happy with the EVA1 (and often GH5s as second camera) shooting 4k / 30p / 10-bit LongGOP 150mbps. Probably well over 100 hours of footage. They don't want to deal with either RAW or LOG, and actually prefer the smaller 150mbps files. I also keep "L" mounting adapters for both cameras in the kit to allow orienting for a vertical frame for head-to-toe single person shots.

                            Comment


                              Originally posted by OnSet View Post

                              Yes, this could have been part of the issue as well, since I was doing VFR 120fps with the EVA1 - which as I recall meant that for full sensor shooting, the sensor would have been set to s35/2.8k recording 1080 4:2:0. I do recall setting up the various camera settings on an sd card so I could quickly switch between 4k 4:2:2 for normal speed, 4k h.265 at vfr 60p, and the 1080 vfr 120p 4:2:0.
                              Of course I'd like to see more options and internal setup files for the EVA2 as well if it ever happens...

                              However -- for normal green screen shoots my corporate clients have been extremely happy with the EVA1 (and often GH5s as second camera) shooting 4k / 30p / 10-bit LongGOP 150mbps. Probably well over 100 hours of footage. They don't want to deal with either RAW or LOG, and actually prefer the smaller 150mbps files. I also keep "L" mounting adapters for both cameras in the kit to allow orienting for a vertical frame for head-to-toe single person shots.
                              Chroma sampling really has no impact on blur or how well something keys. It only impacts pixelated edges really. When we have 4K and 4:4:4 the two chroma channels are 3840x2160. The same as the luma channel. 4:2:2 is 1920x2160 for the two chroma channels and 4:2:0 is 1920x1080 for the two chroma channels. It’s less color detail but nothing else really changes. The images are still the same look. The only thing that changes really is how the edges keyed out can be pixelated or look jagged. Even that is kind of hard to see on 4K. A lot of cinema VFX is done at 2k anyway so really even 4K 4:2:0 has as much chroma detail to work with as most Hollywood movies have had for a decade.

                              if you take 4:4:4 material and convert to 4:2:0 you will essentially pull the exact same keys with the exact same challenges. Only difference is the 4:2:0 may be slightly jagged long the edges. The other thing that can happen is those edge pixels are not just jagged but an average. So a green pixel and a white pixel from a shirt become a pixel 2 pixels wide that are an average of those two colors. This can lead to a very slight extra softness or glow to the edges along with the jaggedness. Not a big deal and you have to be insanely anal and picky to really notice it. Especially at 4K where every pixel is super tiny to begin with.

                              I used to work with DV 4:1:1 with Apple Shake before they killed it and would create my own macros to upscale the chroma channels and pull very accurate keys. The edges were softer than they should have been but it was clean. I do the same thing with 4:2:0 now in Fusion. You split the video into luma and chroma channels, blur the chroma channels and merged the channel back together. Eliminates any jagged edges you could have. When it comes to 4K however you don’t always have to do that and some -refer to just leave the potential jagged edges vs the softer edge detail that may not look as natural. 4:2:2 is even harder to see the missing chroma values.

                              4:4:4 is the best of course but I wouldn’t call it a must. I like 4:4:4 not because of the chroma samples but because it’s not YUV color space and it eliminates any averaged color edges. The 10bit doesn’t hurt either to reduce any fine posterization along subtle green screen values. Again not a must.

                              Comment


                                Originally posted by Zim View Post

                                As long as people can't see it at 200%!!
                                Okay I'll make films and you can enlarge other peoples work to 300%. Keep up the good work!
                                They are both a waste of time. lol

                                ha, I'm just playing...hope it turns out beautifully.

                                Comment

                                Working...
                                X