Ski Touring Film

Started by Welles, June 17, 2018, 09:33:23 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

Welles

Hello,

Here is a short ski touring film, made with raw video on a Canon 100D and a 10-18 Lens. I'd be happy to receive comments on how I can improve and produce a better one next time!

I use the VisionLog profile. Do you think I would benefit from using the Cinelog profiles?



Workflow for each clip:
1. MLVFS -> AE -> ACR : Adjust the image with "Adobe Color" profile, then use VisionLog profile. The project settings are 16 bits per channel, Rec. 709.
2. Put in new HDTV 1080p 25 composition (no upscaling, means I have black borders), export to DNxHD (10 bits)
3. Import in PP, edit in a new 1080p DNxHD sequence, and apply a Lumetri Color effect with "VisionLog to camera neutral Rec. 709" input LUT.
4. Create a new 720p sequence, export to H264.
100D.100B / EF-S 10-18mm IS STM / EF-S 18-55mm IS STM / EF-S 55-250 mm IS II / EF 40mm STM

allemyr

Hi!

How much space does it take to record a minute with your settings right now in camera?

Nice mountain!

Welles

Thanks :)

Once it's in DNxHD format, it's 0.78 minutes per GiB.
100D.100B / EF-S 10-18mm IS STM / EF-S 18-55mm IS STM / EF-S 55-250 mm IS II / EF 40mm STM

histor

Warp Stabilizer plays its own tricks: in the static scene about 3 min. 10 sec. the mountain is breathing. Mocha would be much more controllable.

allemyr

Quote from: Welles on June 18, 2018, 08:13:35 PM
Thanks :)

Once it's in DNxHD format, it's 0.78 minutes per GiB.

Yes but you don't record DNxHD in camera?

Give me some hint? 10 or 14 bit, 720p? Raw?

We al use ML RAW for different reasons, my reason is because that I like fullframe and 14 bit raw. I like grading a lot.
Does the 5D3 enables me to make the most initeresting videos I can make? I don't know but maybe not.

The image that you make has something, but most of the clips are quite similar in grade from an ungraded video you made 4 years ago? Maybe hard to say that.
Maybe a lighter format in camera can make you record more and edit easier?









Took the fredom to screenshot two of your videos.

Well there is so much that can be improved in the video, but I think most of us here only has videomaking as a hobby fun to do thing, and then you can do videos whatever you like :)


Welles

Thank you for your messages, I appreciate your comments.

Quote from: histor on June 18, 2018, 08:53:29 PM
Warp Stabilizer plays its own tricks: in the static scene about 3 min. 10 sec. the mountain is breathing. Mocha would be much more controllable.

Could I get better results by playing with the stabilizer parameters ? I use "position, rotate, scale" as method in this scene.

Quote from: allemyr on June 19, 2018, 03:06:09 PM
Yes but you don't record DNxHD in camera?

Give me some hint? 10 or 14 bit, 720p? Raw?

We al use ML RAW for different reasons, my reason is because that I like fullframe and 14 bit raw. I like grading a lot.
Does the 5D3 enables me to make the most initeresting videos I can make? I don't know but maybe not.

In camera I shot raw video, 25 frames per seconds, at a resolution like 1472 × 616, maybe a tiny bit less. That's how much I get from my Canon 100D. since I did not modify it, I suppose this is 14 bits per channel. I did not find where to set it to 12 or 10 bits yet.

I'm interested in obtaining as much quality as I can. I watched your latest movie , and I can see you like grading, impressive work. I hope to be as good some day.

The video 4 years ago was shot without ML, in H264 1080p.
100D.100B / EF-S 10-18mm IS STM / EF-S 18-55mm IS STM / EF-S 55-250 mm IS II / EF 40mm STM

Danne

https://bitbucket.org/Dannephoto/magic-lantern/downloads/magiclantern-Nightly.2018Jun17.100D101_sd_uhs.zip

Above version and 12bit lossless should give you full resolution in 25fps. Or if you enable sd_uhs you can run it with even higher bits.

PaulHarwood856

Hey Welles,

   Regarding warp stabilizer, read this: https://whoismatt.com/bestwarpstabilizersettings/

   This article really helped me in getting warp stabilizer to work on all shots.

   Also, I used to use Visionlog, but I now use Cinelog-C. I like Cinelog-C, and cannot go back to Visionlog. If you have any questions before purchasing, talk to Andy600 here on the forum.

   I recommend using Batch_mlv if you're on Windows, and Switch (formerly known as CR2HDR). I really like the software Danne provides (I believe there are more people who have worked on the development of this app)

   Danne is really helpful as well. I've asked him questions on this forum, and he is willing to work with you.

   Also, check out this video: https://www.youtube.com/watch?v=VEBVOuh3P3I

   You want your histogram to be as close to 0 as possible when exposing to the right. I wouldn't expose to the point where there are no zebras showing, because the exposure will be off by 2 stops, and the image degrades when pushing up in post.

   Also, try to experiment with your ISOs. If you can find what you native ISO is, then it will be helpful for achieving the best image from your sensor.

   On the Canon 7D I shoot at ISO 800 when possible, if not, then 1600. Especially in H.264, the dynamic range is as best as possible.

   Anyways, I hope this was helpful. Please let me know if there is something I wrote that is confusing. I will do my best to help you with the knowledge I have.

   Cheers!

- Paul Harwood

allemyr

Quote from: PaulHarwood856 on June 20, 2018, 10:30:44 PM
Also, check out this video: https://www.youtube.com/watch?v=VEBVOuh3P3I

You want your histogram to be as close to 0 as possible when exposing to the right. I wouldn't expose to the point where there are no zebras showing, because the exposure will be off by 2 stops, and the image degrades when pushing up in post.

I agree with the person that made that video that you can't use zebras when using raw recording. But I find it quite funny that he doesn't come to the conclusion that thats because ML zebras ofcourse use the videoimage that comes from the display or atleast is the same as the HDMI outbut image, its 8 bit and the camera records 14-bit! Or I do record at 14-bit atleast.





Why did he go with that diagram image? Since when is the limit on raw recording on 5D3 11-bit raw? 11-bit = 2048 10-bit is 1024, and yes 14-bit is 16384 which is quite many more values then 2048. and don't mix this luminence values with RGB or RGB plus Alpha values, 8 bit RGB is 16,7 million colors and 14 bit is 4,4 trillion colors, yes you heard right, 4 398 046 511 104.

Why just use the histogram in the middle? I grade my shadows quite much, and therefore I want an bright image, i overexpose 1 full stop on the histogram. Never a problem to take the highlights back for me with 14 bits.

Well, I think that I really should do a video aswell explaining my point in shooting and post. But that will take time. Maybe time that I should take to make it.

What is Cinelog-C really useful for? I bought it when released 4 years ago, never fit my workflow and doesn't do it now either.

Hers a diagram I made that describes difference with 8 bit and 11-bit, you have to imagine where 14 bit with 16384 luminance values are instead of 8 bit 256.





Ok thats oftopic for the thread, ofcourse but want to clarify

Welles

Thank you so much for your comments, they are very helpful!

Hey Danne,

Quote from: Danne on June 19, 2018, 09:48:40 PM
https://bitbucket.org/Dannephoto/magic-lantern/downloads/magiclantern-Nightly.2018Jun17.100D101_sd_uhs.zip

Above version and 12bit lossless should give you full resolution in 25fps. Or if you enable sd_uhs you can run it with even higher bits.

Thanks for the tip and for your software! I've just tried on camera and I record at full resolution with 12 bits lossless, as you said. I find the user interface is also better.

I'm surprised it's MLV 1.1 though. And unfortunately I get gibberish (pic below) when I mount the files with mlvfs (latest windows x64 build, latest stable dokany), as I do usually. Is it because it is MLV 1 instead of MLV 2? Do I have to process the files with batch_mlv before mounting the mlv files?

Update to this post:
After trying to use mlvfs with 12 bit lossless footage, I wanted to check if everything was OK with my MLV 2.0 files. Unfortunately, no :( I got some green bad/focus pixels back. It remembers me the headaches back in the days, when I was setting it up for the first time.

I tried MLV App, it works. I also tried batch_mlv_in-out, it works too.

I think I'm gonna use batch_mlv_in-out from now on.





Hey Paul,

Quote from: PaulHarwood856 on June 20, 2018, 10:30:44 PM
   I recommend using Batch_mlv if you're on Windows, and Switch (formerly known as CR2HDR). I really like the software Danne provides (I believe there are more people who have worked on the development of this app)

   Also, check out this video: https://www.youtube.com/watch?v=VEBVOuh3P3I

I use Windows indeed (just for photo/video purposes, otherwise I am a Linux kind of guy). I did not know that CR2HDR changed, things go fast...

Your link will definitely changed my way to expose. Until now I was using ETTR with a "no zebra" policy.

I am surprised by the fact that you get more dynamic range at 800 ISO. Is it not that the lower, the better? What is this "native ISO" concept ? Are you saying that there is a sort of "sweet spot", like for lens aperture?
100D.100B / EF-S 10-18mm IS STM / EF-S 18-55mm IS STM / EF-S 55-250 mm IS II / EF 40mm STM

PaulHarwood856

QuoteI use Windows indeed (just for photo/video purposes, otherwise I am a Linux kind of guy). I did not know that CR2HDR changed, things go fast...

Your link will definitely changed my way to expose. Until now I was using ETTR with a "no zebra" policy.

I am surprised by the fact that you get more dynamic range at 800 ISO. Is it not that the lower, the better? What is this "native ISO" concept ? Are you saying that there is a sort of "sweet spot", like for lens aperture?

Hey Welles,

    When shooting H.264 compressed, the dynamic range is the best at ISO 800 on the Canon 7D. When shooting ISO 100, there is less dynamic range. I filmed in my backyard a few months ago, and when I saw ISO 800 had the most dynamic range, I decided to use a Variable ND outside and keep the ISO at 800. My shots have improved with this, combined with ETTR staying as close to 0 as possible. And you're welcome, glad you now know about ETTR vs. Zebras. I also found that f8 gives a really sharp image, however if going for shallow depth of field, I check focus for each shot, because if it's off just the slightest bit, there's no way to sharpen or fix that in post. I like to use a monitor when possible, because while recording I can't zoom in to focus on the Canon screen. Apparently with the Canon C100 (I saw this in a video, don't remember which one), ISO 100 there is less dynamic range, and is more noticeable than with ISO 100 on DSLRs. My point is, try to film a scene, like trees and blue sky with white clouds, and film at different ISOs while at f8. You can also test skin tones at different ISOs. This way you will know the weaknesses and strengths for certain settings on your camera. When shooting raw, the dynamic range might be fine with ISO 100 vs ISO 800 on the Canon 7D, but why not use the sweet spot for H.264 as well as raw video? Please let me know if I can help in any way, and thanks for your response.

- Paul