Hyperlapse Issue

Raptorman · Jun 24, 2020

So I went out today to take some panorama's and hyperlapse's and when I reviewed the video clips DJI makes from the still images of the hyperlapse I was wondering WTF was going on with the focus. Turns out it isn't the focus but that the video is 1080P and I wasn't expecting that. I guess you need to have the still capture set to 48MP to get the 8K hyperlapse, but that didn't occur to me today until after reviewing the 1080P videos. I also found the user interface to setup the hyperlapse to be pretty non-intuitive and couldn't quite figure out how to change the settings to get a longer hyperlapse.

So, a little poking around the internet and I see that the 8K hyerlapse isn't working too well with shaking being the cost common issue, hopefully that will be resolved with a firmware update. I do wish they offered a 4K hyperlapse as an option. Alternatively, if they would permit the stills of the hyperlapse to be recorded so that you can create the hyperlapse in post, that would probably be my preferred option to create a 4K hyperlapse.

Brian

Ice_2k · Jun 24, 2020

Your preferred option is available. Set 1080p mode, enable RAW from the Camera menu and you’ll get the 12MP RAW images to create the 4K hyperlapse from

GadgetGuy · Jun 24, 2020

Why not just shoot a 4K 60fps video, and speed it up in post to your desired speed?

Ice_2k · Jun 24, 2020

GadgetGuy said:
Why not just shoot a 4K 60fps video, and speed it up in post to your desired speed?

Bunch of reasons. Lesser quality, no way to have the drone fly automatically in a smooth pattern, no motion blur which is really nice in timelapses etc

GadgetGuy · Jun 24, 2020

Ice_2k said:
Bunch of reasons. Lesser quality, no way to have the drone fly automatically in a smooth pattern, no motion blur which is really nice in timelapses etc

If the output of both is 4K 120 mbps video, how is it of lesser quality? You can certainly put the drone into tripod mode for smooth patterns and eliminate or create motion blur based upon the shutter speed using ND filters, while shooting 4K at 60fps or 30fps. Just because it is 4K video does not mean it has to be shot any faster than the alternative hyperlapse duration. If the result is going to be a 10 second 4K video, is there really going to be a discernible difference?

Ice_2k · Jun 24, 2020

GadgetGuy said:
If the output of both is 4K 120 mbps video, how is it of lesser quality? You can certainly put the drone into tripod mode for smooth patterns and eliminate or create motion blur based upon the shutter speed using ND filters, while shooting 4K at 60fps or 30fps. Just because it is 4K video does not mean it has to be shot any faster than the alternative hyperlapse duration. If the result is going to be a 10 second 4K video, is there really going to be a discernible difference?

1. The 4K video is compressed, it’s by design going to be off a lesser quality than RAW images. Infinitely easier to recover shadows for instance.
2. Not sure how tripod mode would help. Can you manually maintain constant movement, rotation and ascent? For 20 minutes? While keeping your subject in frame? I find that difficult to believe.
3. ND filters are not enough. Even in 30fps, you can’t go any lower than 1/30. The ideal shutter speed for timelapses, depending on the subject, is around the 1/3, 1/4s mark.

Raptorman · Jun 24, 2020

For what I wanted to do I guess I could have shot 4K video then sped it up in post. The quality may be a tad inferior to a true hyperlapse using higher rez stills, but the difference would not likely be significant and you would have the option to speed up less than the minimum 2s interval in hyperlapse.

I may give it another shot next week.

Brian

GadgetGuy · Jun 24, 2020

Ice_2k said:
1. The 4K video is compressed, it’s by design going to be off a lesser quality than RAW images. Infinitely easier to recover shadows for instance.
2. Not sure how tripod mode would help. Can you manually maintain constant movement, rotation and ascent? For 20 minutes? While keeping your subject in frame? I find that difficult to believe.
3. ND filters are not enough. Even in 30fps, you can’t go any lower than 1/30. The ideal shutter speed for timelapses, depending on the subject, is around the 1/3, 1/4s mark.

If you shooting night scenes, which do make for great hyperlapses with traffic and such, where you want motion blur, I agree. For most daylight hyperlapses, where motion blur is not important, a speeded up 4K video will likely be better than the 1080p result, without needing to manually create the hyperlapse yourself from original images, whether DNG or JPG. The Spotlight capability probably offers the centering of the subject you are referring to, while slowly flying in tripod mode. Certainly, Litchi will eventually offer the type of automated flight control you seek while shooting 4K video.

Ice_2k · Jun 24, 2020

GadgetGuy said:
If you shooting night scenes, which do make for great hyperlapses with traffic and such, where you want motion blur, I agree. For most daylight hyperlapses, where motion blur is not important, a speeded up 4K video will likely be better than the 1080p result, without needing to manually create the hyperlapse yourself from original images, whether DNG or JPG. The Spotlight capability probably offers the centering of the subject you are referring to, while slowly flying in tripod mode. Certainly, Litchi will eventually offer the type of automated flight control you seek while shooting 4K video.

The motion blur is important for day shots as well, otherwise it looks really unnatural with very quick moving objects that are extremely sharp.... but sure, use the 4K video and speed it up if that’s what you want...

boblui · Jun 24, 2020

Raptorman said:
.... I guess I could have shot 4K video then sped it up in post. The quality may be a tad inferior to a true hyperlapse using higher rez stills, but the difference would not likely be significant .....

The difference is not just significant but TREMENDOUS . From my M2P :

Footage synthesized by the drone :

Footage composed from edited RAW stills :

GadgetGuy · Jun 24, 2020

Ice_2k said:
The motion blur is important for day shots as well, otherwise it looks really unnatural with very quick moving objects that are extremely sharp.... but sure, use the 4K video and speed it up if that’s what you want...

Between the 4K 60 fps which already smooths out the video (that's the primary benefit for me) and use of ND filters to achieve twice the frame rate for the shutter speed, jitter shouldn't be a problem for speeded up 4K shot in daylight. However, it just depends upon how much work you want to do for a 10 second clip that likely will get played at a default 360p on a smart phone, unless the user sets up 4k first before playing it, as YT jealousy guards their bandwidth!

GadgetGuy · Jun 24, 2020

P

boblui said:
The difference is not just significant but TREMENDOUS . From my M2P :

Footage synthesized by the drone :

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Footage composed from edited RAW stills :

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Personally, I prefer the contrast in the original, but the M2P is certainly the best drone for this type of night hyperlapse!

Ice_2k · Jun 24, 2020

GadgetGuy said:
Between the 4K 60 fps which already smooths out the video (that's the primary benefit for me) and use of ND filters to achieve twice the frame rate for the shutter speed, jitter shouldn't be a problem for speeded up 4K shot in daylight. However, it just depends upon how much work you want to do for a 10 second clip that likely will get played at a default 360p on a smart phone, unless the user sets up 4k first before playing it, as YT jealousy guards their bandwidth!

You are seriously misunderstanding this... the high 60fps framerate and the smoothness it provides have no bearing on what happens when you speed up that footage to 30x since you will be skipping most of your frames. You’re actually going to be getting the exact opposite effect, with a very rough and jumpy footage. The shutter speed also will be incorrect because that is correlated with the time it passes between frames and you’re changing this by showing one frame and then skipping 29, you’re gonna get much more time passing between adjacent frames. For this example, you’re effectively playing 2 of your original frames in each second, making the ideal shutter speed for those frames 1/4s.

GadgetGuy · Jun 24, 2020

Ice_2k said:
You are seriously misunderstanding this... the high 60fps framerate and the smoothness it provides have no bearing on what happens when you speed up that footage to 30x since you will be skipping most of your frames. You’re actually going to be getting the exact opposite effect, with a very rough and jumpy footage. The shutter speed also will be incorrect because that is correlated with the time it passes between frames and you’re changing this by showing one frame and then skipping 29, you’re gonna get much more time passing between adjacent frames. For this example, you’re effectively playing 2 of your original frames in each second, making the ideal shutter speed for those frames 1/4s.

What you stated makes sense, so, if that is the case, use 24 fps and just fly slower to minimize the movement between the frames not skipped in speeding it up 30x. The bigger issue is your viewer playing a 6 second YT clip on a smartphone. In the two examples above, YT played both at 360p by default on my iPad Mini, which is 3x larger than most people's smartphones, upon which 90% of their media is consumed. By the time YT catches up to the changes made to the playback resolution, the video is already over. I am wondering if the improvement over the default hyperlapse or a 4K speeded up 4K video is lost on the end user, other than the creator, and I am all about quality as a creator, even if I am the only one who can see the difference!

Ice_2k · Jun 24, 2020

GadgetGuy said:
What you stated makes sense, so, if that is the case, use 24 fps and just fly slower to minimize the movement between the frames not skipped in speeding it up 30x. The bigger issue is your viewer playing a 6 second YT clip on a smartphone. In the two examples above, YT played both at 360p by default on my iPad Mini, which is 3x larger than most people's smartphones, upon which 90% of their media is consumed. By the time YT catches up to the changes made to the playback resolution, the video is already over. I am wondering if the improvement over the default hyperlapse or a 4K speeded up 4K video is lost on the end user, other than the creator, and I am all about quality as a creator, even if I am the only one who can see the difference!

By that logic, we can all use a Tello because 720p is enough for most

Bottom line, we already have a mode purposely built for this exact scenario, why are you trying to find a way to minimize the downsides of using a normal speeded up video?.... You won’t be able to get the nice smooth shutter speed no matter what you do and you won’t get the smooth flying the SDK is supported and Litchi enables it.

coneslayer · Jun 24, 2020

GadgetGuy said:
just fly slower to minimize the movement between the frames not skipped in speeding it up 30x

Please tell us your secret for hand-flying for 20 minutes, with aircraft and camera moves that still look smooth when sped up 30x. I admit I'm new, but I sure as hell can't do that.

Ziggy01 · Jun 24, 2020

coneslayer said:
Please tell us your secret for hand-flying for 20 minutes, with aircraft and camera moves that still look smooth when sped up 30x. I admit I'm new, but I sure as hell can't do that.

Lol. Indeed.

Better still, someone show me a time lapse video taken with a drone where one skips 59 frames per 1 that is even vaguely watchable.

boblui · Jun 24, 2020

GadgetGuy said:
Personally, I prefer the contrast in the original, but the M2P is certainly the best drone for this type of night hyperlapse!

The difference is not in the contrast but the amount of information available for editing. The footage from raw stills can be made to look exactly like that out of the drone but the reverse cannot be done.

As the difference is between raw stills and heavily compressed H264/265 video. It is not the privilege of M2P but can be seen on all other models that can store raw stills of hyperlapse.