Personal View site logo
AVCHD 1080p time lapse
  • 67 Replies sorted by
  • @balazer, thank you so much for the explanations.

    For 23.976fps, would I not get the following?

    3 frames per exposure at 1/8s (0.125s <= 3*0.0417s)<br />
    12 frames per exposure at 1/2s (0.5s <= 12*0.0417s)<br />
    I have a mental block to get pass the logic behind the 4 frames/exp for 1/8s and 13 frames/exp for 1/2s. I understood the GOP/I-frame part. My apology for such a bother.
  • This is from Wikipedia..Not sure if this helps but..It explains it much better than I ever could--

    "It is also important to consider the relationship between the frame interval and the exposure time. This relationship essentially controls the amount of motion blur present in each frame and it is, in principle, exactly the same as adjusting the shutter angle on a movie camera. This is also known as "Dragging the shutter."

    A film camera normally records film at twenty four frames per second. During each 24th of a second the film is actually exposed to light for roughly half the time. The rest of the time it is hidden behind the shutter. Thus exposure time for motion picture film is normally calculated to be one 48th of a second (1/48 second, often rounded to 1/50 second). Adjusting the shutter angle on a film camera (if its design allows) can add or reduce the amount of motion blur by changing the amount of time that the film frame is actually exposed to light.
    Blurring vs. exposure times

    In time-lapse photography the camera records images at a specific slow interval such as one frame every thirty seconds (1/30 frame/s). The shutter will be open for some portion of that time. In short exposure time-lapse the film is exposed to light for a normal exposure time over an abnormal frame interval. So for example the camera will be set up to expose a frame for 1/50th of a second every 30 seconds. Such a setup will create the effect of an extremely tight shutter angle giving the resulting film a stop-animation or clay-mation quality.

    In long exposure time-lapse the exposure time will approximate the effects of a normal shutter angle. Normally this means that the exposure time should be half of the frame interval. Thus a 30 second frame interval should be accompanied by a 15 second exposure time to simulate a normal shutter. The resulting film will appear smooth.

    Long exposure time-lapse is less common because it is often difficult to properly expose film at such a long period, especially in daylight situations. A film frame that is exposed for 15 seconds will receive 750 times more light than its 1/50th of a second counterpart. (Thus it will be more than 9 stops over normal exposure.) A scientific grade neutral density filter can be used to alleviate this problem."
  • Thanks, No_SuRRenDeR. I had read those paragraphs and understood the relationship between frame interval and shutter speed for smooth time-lapse. What the paragraphs said was that if the shutter speed was too fast for the frame interval, the time-lapse would be choppy. Did I understand it correctly?

    But what I can't get into my head is the "13 frames per exposure" at 1/2s shutter for 24fps (23.976fps). The fame interval is 0.0417s. For 1/2s shutter, I keep thinking that I will get only 12 frames for the exposure. I am sure I am wrong on this and trying to learn from all of the people in PV.

    Thanks for helping me.
  • @Dusty42, please see my updated explanation, above. http://personal-view.com/talks/discussion/1559/avchd-1080p-time-lapse#Item_25 The old one was all wrong.

    @tida, I think a GOP length of 13 frames (for the 1/2 s shutter setting) is going to be your best bet. You'll get best quality, and really long recording times, like 10 hours in 32 GB. GOP13 here is many times as efficient as GOP1.
  • @balazer, thank you very much for helping me understand.

    10 hrs on a 32G card! Would mind share your settings for GOP13 please?

    Thanks.
  • I took Driftwood Aquamotion v1 and changed the GOP length to 13 frames.
  • Thanks, balazer. I will try that.
  • I know this is different from your discussion on frame rates, but can't think where else to post this: In time lapse videos, sometimes you see the sun travelling across the sky - which means its being focussed on the sensor for a long time. Doesn't this damage the sensor? I'm thinking of what magnifying glasses do to leaves in the sunlight...
  • You should indeed use a strong ND if sun crosses your frame. Otherwise the output of your cam might become very unique afterwards..
  • Is it possible to make intervalled stills with GH2 ?
  • Hi everybody!

    @balazer - You've said: " I would have shot 8-fps in 300% variable mode with a 1/8 s shutter speed".
    What would be the GOP then? 3?

    TIA
  • In Variable Movie Mode, every frame is a new exposure, so you should use a GOP length of 1.
  • @balazer, thank you very much! I didn't know about it.
  • @balazer ok so my question is...Using 1/2s shutter speed with maximum aperture f3.5 on the stock lens 14/42mm whenever i try to get a good exposure using ISO i get way to much dark or too much brightness on the histogram...how can i compensate that when i want to do timelapse...which tricks do you use??? Custom White balance??
  • Do you mean you get too much darkness and too much brightness at the same time? Or it's one or the other? This sounds like basic exposure problems. Just change the aperture and the ISO setting until the picture looks right in the LCD. Don't worry too much about what the histogram shows until you get used to how it works.
  • @balazer I came over here as you mentioned this thread in the FW1.1 topic. I'm using @driftwood 's Quantum patches with a GOP of 1. I don't want to use different patches for different purposes. I'd like to use 24H for regular shots and 24L in variable movie mode that's why I asked @Vitaliy_Kiselev if it's possible to change the value. The variable movie mode is great for timelapsing I think, especially as you get a file with 23.976 individual frames per second. At the moment my problem is that I can't get @driftwood 's latest Quantum v9b to span in 24L. But perhaps I make a mistake and there is a better way? I also tried MJPEG but the quality sucks as soon as I'm shooting 2 fps. But anyway, thanks for pointing me to this topic, very interesting and great results :)

  • @balazer

    Oh yes your are the best!!! thanks ;-)

  • I second the request for modified variable mode to meet the 2 fps. I know there is alot other stuff more important, but it would indeed be nice.

  • @tobnyot, try Aquamotion for variable movie mode.

  • But also without the hack you achieve beautiful timelapse. Just use AVISYNTH for blending and frame jumping. Record at normal 24p or 25p at low shutter speed that you come close to 360 degree. With these recommended 4 blending types of @balazer the advantage is that also noise will be reduced in lowlight at higher iso.

    I use the AVI Script creator of MEGUI and you will have a very short workflow. That is due to the fact that you can use a draft script. From that a new script will be created automtically in the same folder under same name and final mp4 file will be directly generated from the GH2 mts files under same name.... and you can add one jobs after the other which will processed automatically.

  • Following Video is a HDR Exposure Fusion timelapse generated by 1080p 25fps at 1/2s shutter speed, 360 degree and approx 65x times. All out of Avisynth workflow which takes only few minutes.

    Fusion took place by: http://forum.doom9.org/showthread.php?t=152109

    clip1: timelapse video where the clouds were exposed correctly

    clip2: averaged 1 minute video transferred to one single image, which was taken before timelapse started. Where dark areas were exposed corretly.

    It was generated this morning. In south of Germany - only a little bit of snow was fallen on the streets at night. You will find fine details everywhere...

  • For the exposure it´s great, but in thys case it looks strange because the shadows don´t change and also the trees don´t move. Looks like a scary movie.

  • @Kihlian in reality also the trees were not moving by the wind. Movements I have to test - using an averaged picture from a video would have the advantage to erease moving objects.

    But you are also able to work with a clip2 at 1/50s. Then you run at different times. Like cars are passing by as normal and clouds are racing....