Welcome Matrice Pilots!
Join our free DJI Matrice community today!
Sign up

Is ProRes worth it?

I’m posting this lengthy message and video link with hopes that this explanation can help others achieve high quality video with the Inspire 2.

The purpose of this post is to describe and demonstrate a quality comparison between video recorded on the SD card compared to the CineSSD card. Despite an extensive search on the web, I could not find a comparison that truly answered the question of whether or not shooting in ProRes 422 HQ was worth the additional cost and effort.

Background: I’ve been filming this marsh and beach over the last few months typically early in the day or late in the afternoon. With high contrast, the darker or shaded areas always appeared rather muddy or mushy (I do not know the correct term) with the Mavic Pro. I originally thought it was due to compression.

I proceeded to purchase an Inspire 2 with the X4S, and actually noted higher quality yet the muddy areas persisted.

Last month I discovered through a post on the forum that the problem was actually due to noise reduction automatically applied if the sharpness setting was 0 or below. Once I set the sharpness to +1, the video improved considerably on both the Mavic and Inspire 2 with X4S.

However, the quality was still not satisfactory. I therefore purchased an X5S, and although improved over the X4S (with the SD card), the muddy areas persisted to some degree.

Progressing along, therefore I decided to purchase the ProRes license and SSD drive, which I tested this morning.

The difference was extraordinarily apparent viewing in 4K, and clearly visible with compression to 1080P.

My settings were as follows:
Inspire 2 – X5S
SSD: 29.97, ProRes 422HQ, 4K 3840x2160, Rec 709, normal, did not use Dlog
SD: 29.97, H264, 4K 3849x2160, Rec 709, normal, did not us Dlog
The camera recorded to the SSD and the SD simultaneously (didn’t realize that this happens by default).
Sharpening was set at +1 but was only applied to the SD (it does not alter the SSD footage).

Post: FCPX sharpening (2.5) was applied only to the 422 HQ.
The same LUT (CG Falcon Wish) was applied to both videos through Color Finale.
A curves and saturation adjustment was made to the SD video to get it to closely match the mid tones.

Findings: In the attached video I placed the H264 in the superimposed window. In the darker areas, ProRes 422 HQ maintained high levels of detail with no muddiness apparent. Compared to H264, the difference is substantial. Perhaps this is due to 10 bit pixel depth with extended latitude, and markedly little compression for ProRes 422 HQ.

If you proceed frame by frame though areas of the video, you can see the difference even when compressed to 1080P.

Bottom line: if you look close, the difference to me is worth the cost. I realize it might not make a difference for everyone else. Additionally it makes no sense to me to shoot in H264 and then transcode automatically to edit in ProRes utilizing FCPX.

From a practical perspective, the CineSSD is fast with a transfer this morning of roughly 140GB in about 7 minutes. FCPX ingests and plays the video without a hiccup. This workflow to me makes more sense. The only minor downside is that the ProRes HQ video is not rectilinear (easily fixed with FCPX).

Hope this helps anyone who is trying to determine if ProRes is worth the extra cost and steps.

While the 4K version is impressive, consider downloading this dropbox link below. I suggest downloading the file rather than playing it online for better quality.

Dropbox - Beach Test422.m4v
 
The compression artifacts - shadows mushiness in particular - are pretty obvious on UHD videos taken with X5 camera. That's what you can expect from 60 mbps technology. The issue, however, seemed to disappear with 100 mbps generation cameras on ridiculously affordable Phantom 4 Pro. One may assume that similar H264 picture quality will be delivered by much more sophisticated - and expensive - X5S camera, leaving SSD and ProRes lossless technology for the most demanding applications. I beg you to correct me if I'm wrong.
 
Dobmatt

The video was shot with the X5S. The bottom line is that you can expect significant improvements as the compression rate drops. The SSD footage in Prores 422 is far less compressed and as a result, the file size is substantially larger.

In a well lit scene without a lot of shadows, it would not be as noticeable.

Ultimately, however, the difference in 4K on a large monitor or TV is obvious.

Thanks for your post.

Barry
 
  • Like
Reactions: John Bishop
Dobmatt

The video was shot with the X5S. The bottom line is that you can expect significant improvements as the compression rate drops. The SSD footage in Prores 422 is far less compressed and as a result, the file size is substantially larger.

In a well lit scene without a lot of shadows, it would not be as noticeable.

Ultimately, however, the difference in 4K on a large monitor or TV is obvious.

Thanks for your post.

Barry

Barry: in your original post, you mentioned the ProRes footage is not rectilinear (which I noticed as well). Just curious what you're doing to correct that. Do you use the Fisheye effect in FCPX, or some other solution? I tried Lens Undistort but it seems tailored more to GoPro cameras. Thanks for sharing your findings!
 
Gamete

I’ve tried several plugins. None of them worked. The fisheye effect in a FCPX is by far the best.

I shoot 5k and that’s the only thing that really works. With the others I would get a bimodal dip.

Barry
 
Dobmatt

The video was shot with the X5S. The bottom line is that you can expect significant improvements as the compression rate drops. The SSD footage in Prores 422 is far less compressed and as a result, the file size is substantially larger.

In a well lit scene without a lot of shadows, it would not be as noticeable.

Ultimately, however, the difference in 4K on a large monitor or TV is obvious.

Thanks for your post.

Barry
I believe that affordable aerial systems finally reached quality levels where average human can not tell the difference between 4K filmed in H264 and Prores. It reminds me similar scenario in audio technology a decade ago, when recording industry tried to utilize the advantages of 96KHz or even 192KHz sampling frequency only to deliver final product in standard 44.1KHz CD format. HD video is still a bit rough for some purists like myself, but 4K UHD is very close to what I can see on stills taken by decent DSLR camera. With introduction of higher recording bandwidth in X5R and X5S cameras there's no need for much more in this class.
 
Gamete

I’ve tried several plugins. None of them worked. The fisheye effect in a FCPX is by far the best.

I shoot 5k and that’s the only thing that really works. With the others I would get a bimodal dip.

Barry

Thanks! I think I’ll take some inside shots and mess around with the settings to see what I can come up with.

Probably need to do this for all the different lenses too since they’ll likely have different amounts of barrel distortion.
 
Compare h264 and prores on inspire 2 x5s, 15mm lens

I actually like the h264 on the right better than the Prores on the left which appears to fall victim to the awful magenta tint in the DJI cameras. The sky in Prores looks "burple" and not anywhere close to the color of the sky on the right which looks more natural on my screen and looking out the window here for comparison. DJI really needs to fix their burple cameras with a tint control, or keep making expensive toy cameras.
 
I actually like the h264 on the right better than the Prores on the left which appears to fall victim to the awful magenta tint in the DJI cameras. The sky in Prores looks "burple" and not anywhere close to the color of the sky on the right which looks more natural on my screen and looking out the window here for comparison. DJI really needs to fix their burple cameras with a tint control, or keep making expensive toy cameras.

I think he shot that with an order version of the fw. The latest version .300 fixed the WB and tint issues (or at least it’s very good in AWB). They just need to make the AWB setting match the same tint when manually setting the temp to same value.
 
UT,
I too prefer the h.264 video. What can you a tribute this difference to? All post processing the same. A post above suggests an older firmware version.

You are an award winning cinematographer, please share with us your opinion on how someone can get the superb h.264 quality without spending the extra budget on license and SSD costs.

Thank you for posting this video demonstration.
 
  • Like
Reactions: Ftolino

Members online

No members online now.

Forum statistics

Threads
2,727
Messages
25,387
Members
5,600
Latest member
RomanChrz