Hi,

Is it possible to get useable still shots from video footage?

Many thanks.

The VLC app on PC has an option to take screenshot of a video you are watching.

It's possible, but the frame grabs won't be as good as those you'd get shooting stills.
If you use ND filters to slow the shutter speed, individual frames might be very poor quality.

Yes. I do it all the time. Some satisfactory results. Not as good perhaps as a still image, but pleasing nontheless.

yes i use wideopad to pick a frame

high shutter speed and at least 60fps will work.

Hello there Hunter_UK. Good day and thank you for reaching out. I hope that our fellow DJI Co Pilots can give out what 3rd party application / software they are using to get useable still shots from video footage. Thank you.

If you are on a PC install Irfanvies and its plugins. You can use it to extract all the frames to a folder and choose the best ones.

I use mpc-hc with Alt-i You can grab any frame with original quality 4K = 8Mp not bad ;)

You can use the capturing button from Premiere Pro.
Each set of frames in each second have at least a Master Frame that is not compressed (the reference frame). If you know how to find it, you grab it and the result is good.

Can do the same in Openshot Video editor too

Pause it on the exact frame you want and click the "camera" icon

Yes. I do it also all the time. Best results for me, film with 60 frames 4k, playback full screen with VLC and take a snapshot as png. Afterwards edit with lightroom

Geo_Drone Posted at 7-24 23:46
You can use the capturing button from Premiere Pro.
Each set of frames in each second have at least a Master Frame that is not compressed (the reference frame). If you know how to find it, you grab it and the result is good.

How do you "find it"?  Is there some marker in Premiere?

Burstmode Posted at 7-27 14:39
How do you "find it"?  Is there some marker in Premiere?

You can go frame by frame and you will see one with a lot less artifacts than others...At first can be a little hard, but in time you can distinguish it very fast.
Look with zoom at areas with very close colors and shades, you will see "larger pixels" (have no idea how to translate this in English) in intermediate frames, but one frame will be visibly better and without "artefacts". That is the Master Frame. You will see one if you look in 8 frames (consecutive frames).
(sorry, my technical english is not advanced).
Cheers.

Geo_Drone Posted at 7-28 00:41
You can go frame by frame and you will see one with a lot less artifacts than others...At first can be a little hard, but in time you can distinguish it very fast.
Look with zoom at areas with very close colors and shades, you will see "larger pixels" (have no idea how to translate this in English) in intermediate frames, but one frame will be visibly better and without "artefacts". That is the Master Frame. You will see one if you look in 8 frames (consecutive frames).
(sorry, my technical english is not advanced).

Thank you.  This actually sounds a bit tedious (lots of work).

Do I understand correctly that ONE (literally, exactly one) frame each second is the "master" frame -- no matter what video settings you are using, there is always a single "maximum quality" frame each second that's used for software interpolation?


The Mini 3 pro specs says it can shoot at 24/25/30/48/50/60 frames per second.  Assuming you're shooting at 60 fps, does that mean that you need to (visually) search through 60 frames, looking for the single frame that looks better/clearer?  (I believe the Mini 3 Pro can also shoot at 120 frames per second at a lower resolution for slow-motion...in that case you would be looking through 120 rames to find the "good" one.)

When looking at individual frames or images, I can easily recognize software artifacts (for example, from compression or upsampling or just processing noise).  If you look close enough, there's always some of that.  I think what you're saying is that there should always be one frame every second of elapsed time (so 1 out of every 60 at 60fps, 1 out of every 24 at 24 fps) that clearly looks "better" with more real capturd data, while the other frames all show some (visibly more) artifacts due to software interpolation ("filling in" the space with data that has been processed more).

Or do you see the "good" frame every 8 frames, no matter what frame  rate you choose (because, I would guess, the actual capture rate is 8  frames per second, and all the higher frame rates are achieved through  software interpolation)?  Searching through 8 frames would certainly be  less work that searching through 120 frames...



Is all that roughly correct?  Thanks!

PS: I have not tried this myself (old computer, have Premiere Pro but  it's painfully slow; have a new very fast computer just delivered --  still in the box! -- have not set it up yet, but will be using Premiere  happily on that one in a few days).

PPS: Your English is pefectly good, thank you for using my native language (we would not get very far if I was writiing in Romanian -- Mulțumesc!)

Usable in what sense?

Showing someone a screen grab of a video - yes.
Posting it to IG - Borderline yes, depending on how well it's been shot.
Editing it for showcasing your skills as a photographer - absolutely not.

For social media, most certainly,
for sending to a print shop to go on a large canvas, doubtful.

To get the best results, film in 4K.
It will save the image at the size of the video, e.g, 4k will save a 3840x2160 still, 960x540, size 5.5mb and it will save an HD (960) at 960x540, size 230kb, just examples but you should get the idea.

DreamWorks Posted at 7-28 18:49
For social media, most certainly,
for sending to a print shop to go on a large canvas, doubtful.

What are you talking about?
No-one shoots anything at 960x540.

Labroides Posted at 7-28 20:34
What are you talking about?
No-one shoots anything at 960x540.

As I said, it was just an example, nothing more, it was, however, based on the screen recording from the DJI RC

DreamWorks Posted at 7-28 21:19
As I said, it was just an example, nothing more, it was, however, based on the screen recording from the DJI RC

Not sure what it was an example of.
No-one shoots anything at 960x540.

Labroides Posted at 7-28 21:42
Not sure what it was an example of.
No-one shoots anything at 960x540.

Example,

Person creates a screen recording on the DJI RC, person takes a still from said screen recording. Still image is 960x.....

DreamWorks Posted at 7-28 21:47
Example,

Person creates a screen recording on the DJI RC, person takes a still from said screen recording. Still image is 960x.....

A 960 x 540 still is no use to anyone.

Labroides Posted at 7-28 21:53
A 960 x 540 still is no use to anyone.

I didnt say it was, it was just an example of the image size from that video size, that's all, why the issue?

Nothing was specified in the original post, the OP didn't even say they were shooting video, just asked about useable still shots from video.

Once again, it was just an example of what one may come across when using the Mini 3 Pro.

The screen recording is a video, isn't it? screenshots may be needed from that video, so why not mention that size? but once more, it was just an example, its not like i was saying that's how the Mini 3 Pro shoots video, is it?

Burstmode Posted at 7-28 09:34
Thank you.  This actually sounds a bit tedious (lots of work).

Do I understand correctly that ONE (literally, exactly one) frame each second is the "master" frame -- no matter what video settings you are using, there is always a single "maximum quality" frame each second that's used for software interpolation?

Multumesc to you too
Does not matter if you use 24-30-60 FPS, is about how HEVC codec works: is taking one master image each 8 frames, is building "blocks" from that image (practically is splitting it in parts) then next frames records only changes in areas that are really changed. In rest, is keeping the area from master frame...this is how HEVC compression works, reduced at a simple explanation (is complicated, but this is the basic of video coding in HEVC).
This is why you have that Master Frame each 8 frames, no matter what FPS you use, the codec is the same. Of course, recording in H.264 changes things, but most of us use H.265.

To answer to other people here that say the image is limited: Well, depends...If you record in 5K and have a master frame saved in PNG from Premiere, edited properly in Photoshop and magnified with Topaz Gigapixel, will result an image that can compete with a photo taken by a normal camera (not mirrorless or DSLR, OK? as not even a DNG from Mavic 3 gets close to a Mirrorless photo, no matter is having 4/3 sensor, optics sucks).

So practically depends on your editing skills...
One more thing: If you import and edit the video then export it again in video, even if is HEVC, is a high possibility to rewrite the master frame at reencoding, meaning that you will not get a real master frame generated at first encoding from camera. Because of this, my work is like this: import video in Premiere, find Master Frame, edit video (WB, colors, contrast and so on), then capture the Master Frame edited and save to PNG. In this way I get maximum quality....If you reencode, the codec will re-arrange the Master frames as you can make changes in FPS (and frames are eliminated or added), so your true Master Frame is gone, will be a replacement. Is hard to notice differences but it is there (artefacts from compression, color shifts, harsh edges...). You can preserve at encoding the real Master Frames but this is already not a beginner tutorial

Cheers.