Testing Danne's new EOS-M ML Build (7/29/2020)

[SoFloCineFile]

Hello everyone, 

I just recently picked up an EOS-M after being inspired by Zeke and wanted to share a short film I shot using Danne's new ML build (7/29/2020): https://bitbucket.org/Dannephoto/magic-lantern_jip_hop_git/downloads/crop_rec_4k_mlv_snd_raw_only_2020Jul29.EOSM202.zip

YouTube link to my short film, "A Venetian Splendor on the Gulf:" 

 

 

Shooting Conditions:

Shot hand-held using neck-strap to stabilize

15-45mm EF-M lens

23.972 fps

12-bit, raw 4k anamorphic and 10-bit 2.5k raw clips

Exported from MLV as h264 12-bit mov, up-scaled to 2160p and graded in LumaFusion

[rawshooter]

Sorry, but this looks pretty terrible (blurry and grainy). Maybe something went wrong either during the shoot or postprocessing?

[SoFloCineFile]

24 minutes ago, rawshooter said:

Sorry, but this looks pretty terrible (blurry and grainy). Maybe something went wrong either during the shoot or postprocessing?

I'm still learning and found out that my editing workflow was not ideal. I was using Zeke's suggestion of exporting the original MLV Raw footage to H265 MP4 and then doing all the color grading and editing on my iPad Pro in LumaFusion. The reason I'm using this workflow is because my desktop is really old and encoding even 2k video is painfully slow. With LumaFusion on my iPad Pro however, I can edit seamlessly and encode 4k almost in real-time.

Did you play it back using 1440p and 2160p on YouTube? I didn't think it was that bad honestly when you play it in those higher res modes on YT. I played it back my 4k monitor and did notice some grain on the higher ISO shots...I used the 15-45mm ef-m kit lens) which is not the best for low light btw...That being said, I don't see such a glaring difference as you describe when comparing it with other EOS-M videos shot under similar conditions and using the 15-45mm kit lens. If you would be so kind as to give me a reference point for what you would consider a non-grainy image, I would really appreciate it. 

[KnightsFan]

I played at 1440p and I have to agree, it's very noisy and not sharp. When you watch the version on YouTube, how does it compare to your H265 intermediates? How about to your original MLV files, if you have a way to play those back? I haven't watch any other EOS M footage so I can't say whether this is typical of that camera, but I checked against the XT-3 footage I'm editing right now (which is on a 1080p timeline) and it's night and day the clarity and noise level--not to force a comparison but just to make sure I wasn't going in without an immediate reference.

In any case, you've got some great stuff to film and practice on at that museum! Tinkering with ML was always fun.

[SoFloCineFile]

I will go back and encode directly from the lossless DNG files created in MLV app and upload that version to YT. That being said, if your point of reference is a Fuji XT-3, that's not an apples to apples comparison based on the respective specs of the two cameras: https://cameradecision.com/compare/Canon-EOS-M-vs-Fujifilm-X-T3

Really the only similarity between them is that they are both mirrorless...other than that, the Fuji is 6 years younger, 5 times more expensive ($1000 vs $200), has 8 more megapixels (26 vs. 18 on the EOS-M), and also has a BSI-CMOS sensor vs. a standard CMOS sensor. Just from those specs, it seems obvious that the EOS-M would have more noise, especially in low light, and less sharpness.  Also looking back at my shots, I had to push the ISO to 3200 in some scenes, so keep that in mind as well. My 15-45mm kit lens has a minimum f-stop of 3.5...so not exactly the best for low-light. 

I hope Zeke can weigh in on this issue as well, as he has way more experience than I do in terms of pushing the EOS-M to its maximum possible image quality, but this seems like it has less to do with editing/encoding and more to do with the fact that your camera just has a better image sensor. 

[KnightsFan]

Yeah, definitely not an apples to apples. It was just what I had open at the moment.

[SoFloCineFile]

Updated video using higher quality export settings workflow:

RAW MLV clips from EOS-M ---> CinemaDNG lossless ---> Edited and Graded in Davinci Resolve 16 (free version) --->Final delivery codec for YouTube Export was .mov / PhotoJPEG w/ 250mbps bitrate.

Still some noise due to high iso settings (3200+ for the indoor shots) but better than initial export above (h264 codec)

 

[Thomas Hill]

That looks really good on my phone. What's the max clip length in 2.5k raw?

[SoFloCineFile]

1 hour ago, Thomas Hill said:

That looks really good on my phone. What's the max clip length in 2.5k raw?

Recording about 1 min of 2.5k RAW produces a file size of about 4 GB... I'm currently using a 128 GB Sandisk Extreme Pro (170 MB/s bitrate) and so I should be able to record about 30 min.

That being said, I'm still learning Magic Lantern and have not had a reason yet to record more than a few min at a time. As soon as I get a chance though, I'm going to do a test recording and see exactly how long the camera can actually record in raw without stopping. Will post a screenshot of my results 🙂

[Thomas Hill]

2 hours ago, SoFloCineFile said:

As soon as I get a chance though, I'm going to do a test recording and see exactly how long the camera can actually record in raw without stopping. Will post a screenshot of my results 🙂

Cool, I was wondering because I've read that some of the probably older builds could only record a few seconds/clip.

[SoFloCineFile]

I can confirm that the Magic Lantern build I am using does bypass the factory record limit on the Canon EOS M!

I was able to record continuous 2.5k Raw (2512 x 1190, 16:9 format) for 33:51 minutes, reaching the capacity limit of my 128 SanDisk Extreme Pro (170MB/s).

See snapshots below: 

[SoFloCineFile]

4 minutes ago, Thomas Hill said:

Cool, I was wondering because I've read that some of the probably older builds could only record a few seconds/clip.

Here is the latest build which I was using for the recording test above:

https://drive.google.com/file/d/144TxEEVYpHpMK0t77y3sKj_ZeTGRuY3L/view

[SoFloCineFile]

17 minutes ago, Thomas Hill said:

Cool, I was wondering because I've read that some of the probably older builds could only record a few seconds/clip.

This has turned out to be an amazing camera and the best $150 I've ever spent! I purchased it pre-owned a few weeks ago just wanting something i could quickly use on the go for quality shots and now I have a cinematographic beast on my hands thanks to this latest ML build!

[KnightsFan]

It looks so much better now, so it was indeed the workflow that was causing issues. There's still some noise in low light, but the mushiness is gone and the color has life in it. Also, I'm not sure if you slowed it down before or had a frame rate mismatch, but that opening shot of the front gate used to have jerky and unnatural movement but looks normal now.

One thing though is that you've got some over exposure that wasn't present before. You might want to adjust the curves in Resolve, or experiment with the raw settings. The shot of the bust at 0:44 in the original is properly exposed, but that shots at 0:48 of the bust in the new one is blown out. But yeah, definitely looks like it was captured in Raw now, which the first video did not.

[SoFloCineFile]

15 hours ago, KnightsFan said:

It looks so much better now, so it was indeed the workflow that was causing issues. There's still some noise in low light, but the mushiness is gone and the color has life in it. Also, I'm not sure if you slowed it down before or had a frame rate mismatch, but that opening shot of the front gate used to have jerky and unnatural movement but looks normal now.

One thing though is that you've got some over exposure that wasn't present before. You might want to adjust the curves in Resolve, or experiment with the raw settings. The shot of the bust at 0:44 in the original is properly exposed, but that shots at 0:48 of the bust in the new one is blown out. But yeah, definitely looks like it was captured in Raw now, which the first video did not.

Thx for that. I just realized how to make scopes visible in Davinci...did not have them visible while editing this and also realized that I had kept my brightness low on my laptop--so couldn't tell just by looking at it that the scene. you are mentioning was overexposed.

[SoFloCineFile]

Also, just want to give a disclaimer that I made a mistake earlier when I said that the 33 min video test recording mentioned above was 16 by 9 format. I opened the recording up in MLV and its actually 21 by 10...however, that's still a fuller look than 2:35 or 2:39, so it's awesome imo to be able shoot continuously in this mode, as well as 2.7k and 2.8k raw continous in the cropped aspect ratios.

Before this latest build, I was only able to get continuous in the 1080 rewire and anamorphic pixel binned modes, where the image needed to be stretched in post and didn't look like true 2.5k.

My next goal is to do a short film and try to film it entirely in 2.5k or above resolution.

[aaa123jc]

This is interesting. 

I'm always amazed by what the EOS M and magic lantern can do. I love mine very much and it still is one of my favorite mirrorless camera. Although I will not take it to paid jobs, it is very fun to use and despite the price, has so much to offer. Shooting RAW continuously in such a small body is simply crazy. Though I had never thought of trying it, for some reason, and I only used magic lantern for focus peaking and audio level monitoring. 

Anyway, I suggest to invest a prime lens because the kit lens really isn't that great. Many of the shoots in the video would benefit from a shallower DOF and lower iso setting.