My experience with 5D3 MLV and VisionLOG

Started by ChadMuffin, November 01, 2014, 02:56:34 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

ChadMuffin

Here is a project that I recently worked on using Magic Lantern MLV and VisionLOG and their LUTs, specifically their M31 LUT. This is not a review, critique or anything like that of the tools used, ML or VisionLOG. This is me explaining my experience with them. The reason I am doing this is to show new users of Magic Lantern my method, which not be the best for you, and for the more seasoned users to see other workflows. As a disclaimer, I would not classify myself as an expert. I am completely self-taught in the sense of never attending professional classes. I have just used online sources and my own experimentation to develop myself as a filmmaker. Here is a list of the gear I used in this project: Canon 5D Mark III, Canon 50mm f/1.4, Tamron 70-200mm f/2.8, Tamron 24-70mm f/2.8, Komputerbay 64GB x1000 x2, Tripod (for interviews only), Monopod (ever other shot), LED lights x2, Zoom H4N and Sony UWP-V1 wireless lav.

Now that's out of the way, here is my experience.

The producer requested to have a unique look to the film and I showed him clips from the M31 LUT which has been popular nowadays and he wanted that look. He wanted to show the event to promote local music in St. Cloud, MN and wanted to film the live show, with the audio tracks from the bank and to cut in some interviews of the bands and this was the result.

I recorded the first and last two songs of each of the two sets because those are typically the most energetic and in between I filmed the interviews. I knew there wasn't going to be too much footage so shooting MLVs would be reasonable. Sometimes the cards I have don't record continuous at 1080p so I filmed in 2:35:1 to save space and to ensure I did not have any dropped frames. When my cards filled up I offloaded them on an external hard drive that has USB3.0, formatted and put it back in my pocket ready for the next time I needed. This was a fast paced, informal shoot. When shooting the band and the other non-interview shots, I only used my monopod and did my best to keep it stable and used warp stabilizer in post to fix any little shakes I could have had. Most of the band shots were at 6400 which was handled very well by MLVs compared to my previous attempts with Canon's H.264.

When I got home for the event, I converted all of the MLVs to DNGs with Chmee's raw2cdng. I let that go over night and when I woke up I used After Effects (in 16 bit) to convert them to DNxHD 10bit 175 with VisionLOG. This is a slow process but, I did not have a set timeline for this project to be done so that was okay. If I had a tight schedule, this probably would not be the best option. The only issue I had with this process is I think I had some hot pixels. On some of the 6400+ ISO shots, there were white dots that would come in and out of the individual frames. I think this issue will be fixed when Chroma Smoothing is introduced into the raw2cdng too. But, it is not the end of the world for me and this project. I'm sure Youtube compression will fix this issue.

I created four sets of presets in ACR to make the process faster. The first one I had set the basic settings I wanted to start out with. It would clear any other setting back to 0 and it would include WB changes in increments of Tungsten, Sunny, Shade, ect. The second would then include lens corrections for my lenses, the third would be noise reduction with low, medium, and high. I would then review the image to make sure it looks how I want it and then I would place the last profile with VisionLOG. This process was quite fast for me and I was happy with it and all the control I have. For the interviews I included the .wav file but for the other clips, I did not include audio since I would not be using it, and it would save a lot of clicks for me. I created new compositions for each clip, saved, and exported in Media Encoder and let it do its thing. I then came back later that day and edited it just like normal in Premiere Pro with the VisionLOG LUTs, Magic Bullet Looks and other color grading plugins included with the software.

All in all, changing MLVs to DNGs and then from DNGs and DNxHD had me behind the computer for additional 30 min. The actual converting process added about 16 hrs of the computer thinking when I wasn't around to this relatively short project.

I would like to hear from you of the process that I had. I would love some insight if others have had better results with other options. I did choose to stay in Adobe products because of the dynamic link feature since I also do some audio work in Audition. I also use After Effects and Photoshop often in my video work. It just is easier and more comfortable for me to stay all in Adobe.


NitromanX

Great post and great video footage. Thanks for sharing (i only just seen it!) :)

I'm just experimenting with VisionLog ACR Profile. I see you changed several settings and then applied VisionLog profile at the end. Did you get any flicker from using ACR ? I guess since you didn't tweak the highlights and contrast etc that there were no problems ?

Also, what colour profile do you actually shoot in on camera - Standard / Portrait / Neutral etc ? I'd like to know how to get a camera profile that actually looks like the final footage so i can see the contrast etc.

When do you apply sharpening ? In ACR / in output or / in post ?

I'm a bit of a newbie so your suggestions and thoughts are of great interest.

Thanks :)

g3gg0

Help us with datasheets - Help us with register dumps
magic lantern: 1Magic9991E1eWbGvrsx186GovYCXFbppY, server expenses: [email protected]
ONLY donate for things we have done, not for things you expect!


DeafEyeJedi

Indeed, very nice and perhaps you can try using @Andy600's Cinelog-C (which works well in ACR within AE) and personally I think it comes out better than VisionLog (I own both) but that's jm2c.

Great work & Thanks for sharing @ChadMuffin!
5D3.113 | 5D3.123 | EOSM.203 | 7D.203 | 70D.112 | 100D.101 | EOSM2.* | 50D.109

theartofweb

I think 16 hours for the converting process is a lot. Also, I don't see the use of DNxHD which renders everything to its own chosen template dimensions. Yet, if it is crucial sto stay on adobe there's no other workflow I can think of.

My view is you don't really cut the raw, you convert it to Pro Res or something that you can edit just fine, go back to the raw to color, then master. So, I'd import RAW to Resolve, create proxy for editing, edit proxy, send edit to resolve, grade RAW.

DeafEyeJedi

That's when Life becomes eAsier while using @dmilligan's MLVFS with its virtual files (without taking up Disk space) and has real-time playback in PP on Mac running Yosemite.

Happy Mac. Happy Life!
5D3.113 | 5D3.123 | EOSM.203 | 7D.203 | 70D.112 | 100D.101 | EOSM2.* | 50D.109

theartofweb

I was wondering when it comes to those BMD Film to log luts, if I should use the Log Controls in Resolve instead fo the Primaries; based on the film timing approach, in that, to use a combination of exposure implemented as offset in Resolve and contrast to alter the color balance and overall level on the image. Can the Cinelog-C be compared to the standard Cineon type log curve?

I am asking this because I believe the Log Controls in Resolve may have been designed to be used with standard Cineon type log curves.

Thanks


ChadMuffin

Thanks for the input!

@NitromanX - I did not notice any flickering in the footage. I did not adjust any setting that I have found online to be adaptive, or change frame to frame. This includes Blacks, Whites, Highlights, Shadows, Contrast and Clarity. The Exposure, WB, HLS, Curves, NR, Lens Corrections are not adaptive and are okay to adjust. The grain setting does not move image to image so do not apply this.

I shot in Cinestyle as a picture style but, this is not required for RAW. A picture style is not embedded into the image when shooting RAW. I shot in Cinestyle to more properly judge my exposure at a quick glance while recording. But, this is something I like to do, no need for it.

As for sharpening, and this might be the wrong way to do it, I applied in ACR when I exported the cDNGs to DNxHD. I also applied noise reduction in this step to reduce the process time while editing and the final output.

@DeafEyeJedi - I am sure there are many benefits for using Cinelog-C compared to VisionLOG. But, my business is still growing and I do not have the budget to change my workflow in that way. Interested, but not at this time.

MLVFS would be great but, still being developed for Windows. No beef with Mac, just the cost.

@theartofweb - That is correct, I did not edit the RAW footage itself. Integrating Resolve would make working with the cDNG files easier but, I do not want to go back and forth between different applications as I tend to use After Effects and Photoshop in my edits and using the Dynamic Link makes it so much easier for me. I want real time updates with the main timeline in Premiere when I make any changes in Audition, Photoshop and After Effects. I guess it is a method to my madness and I constantly go back and change things throughout the edit

Because Premiere does not handle ML RAW the best (yet), I wanted to get the footage close, correct WB, then export it with a flat profile in a high bit-rate and bit depth file. DNxHD 175 10-bit worked great for me as I am on Windows and do not have ProRes readily available. To be honest, I am hoping for 'real' adjustments to RAW in Premiere/Speedgrade just as if I were to use Resolve but, who knows if those days will ever come. Maybe there is but, I guess I don't know enough about it.

Licaon_Kter


Quote from: ChadMuffin on April 29, 2015, 10:38:54 PM
MLVFS would be great but, still being developed for Windows. No beef with Mac, just the cost.
You can set up a Linux box (or a virtual machine) that just loads the .MLV via mlvfs and shares the out folder via SMB for like zero$ (yeah yeah time is money)

ChadMuffin

Interesting concept. Thanks for sharing!