Messages

I use 2 display presets. I keep one clean and the second one is where i have histogram, waveform and level indicator. When i record i use the empty one. I use digic peaking, should i turn it off?
I was under the impression that 3.5K is 25fps and 3.3K is 24fps. So i use 25fps in the 3.5k preset. I shoot at 12bit and 16:9 ratio set in the movie tab menu.
My cards are 32gb lexar x1000 and sandisk extreme pro 64gb. No external monitor atm. To achieve continuous recording i lower resolution in the raw recording menu to 2880.
Changing the registers by -10 seems to fix the problem(recording at 2880X1620). Need to test more to be certain.
I even shot a dual iso clip at 1620, no issues.

Can the change in the register be saved in the manual config? eg. CM3 set to 3.5k with register's change?
Another issue that sometimes happen is that the screen gets stuck in real-time focus mode(in 3.5K mode). I cannot change back to framing from ml menu or by going to canon menu and back again. Need to either restart or switch to review and back. I haven't found the steps to reproduce being stuck in real-time yet. Will look more into it.

Thanks for the really quick fix!
Happy to help with the bug hunt. Least i can do for now.

I tried working with 3.5k x5 centered but it's not as robust as 3.3k. I always get compression error around 1 minute mark no matter what resolution i use. Console comes up and i have a few logs. The cam keeps recording but i think a few frames are gone. Is this a known issue, performance related? I can look into in a bit more if you want(upload logs, clip)
I guess one more frame per second is really heavy on card speed compared to 3.3K.

I tried resetting both canon and ml settings. Formatted both cf and sd and installed vanilla 28feb build but still the same issue. It's on M mode. It's not readjusting iso or aperture. Shutter number changes as it should. There is no difference though. Happened both on manual and electronic lenses. Other than that i cannot think of anything else to try.
Tried to change shutter in shutter priority mode. As shutter speed decreased 1/4000, iso maxed and the image got brighter.

Last think left to check is 1.1.3 version, although i prefer 1.2.3 for hdmi monitoring.

With February builds, shutter speed doesn't affect exposure when in video mode 3.5k or 3.3k. The numbers do change but there is no effect on the image or histogram. All other modes work as they should. When i switch to photo the exposure is correct. I tried changing shutter range, shutter lock(changing shutter from ml only), ettr, expo override but none seems to be able to change the actual shutter speed. The number changes, the exposure doesn't. I am on 5D 1.2.3

Is that a bug or is there any other option i am missing?

With the latest versions of danne's ml for eos-m, I can't find how to switch off modules.

I have 2 sandisk cards and 1 sony. The sony doesn't work with sd hack, it's speed is cut down by half(40mb/s to 20mb/s).
40mb is okay for 1080p rewired though.

Did i miss something? Is the toggle removed and is it easy to add it back again?

It's the same in photo and video mode, 0x1f80000

Anybody else has problems with timecode?
When i import audio and dng sequences in davinci resolve the timecode of audio and video is different by a minute, maybe more maybe less, which makes syncing tedious because i have to fix every audio file timecode manually.

Nobody has this problem? I think 1.5something was working ok on that regard.

Still, latest versions are very fast and stable!

Nice idea rainless!

I did my thesis on color grading, the basic stuff(balancing, correcting, applying looks and so on)
It can be very hard to go on depth about color grading without needing a few hundreds pages of analysis.

Your primary notion about color grading(balancing first and then applying looks) is spot on, although i wouldn't trust my eyes when it comes to correcting, especially when working at a park with a laptop(most laptop displays are not even decent, mine's at least is not).

So i ll suggest a few books that helped me, (you seem experienced cinema guy but in case they do help you)

Interactions of color, Josef Albers,
This book is about design in general, but it really focuses on how our perception make colors interact and create illusions. Main reason you need a dedicated place for color correction

The Visual Story, Bruce Block,
this is a well-known book, it's about cinematography but it covers stuff like color pretty well.

Also Alexis van Hurkman books are generally very helpful.
You may also want to check out a few websites on color and perception but maybe it's not so necessary for your goals(i can give you links if you like)

As for shooting raw, i think most professional photographers don't take into account simple stuff such as ettr.
I have learned the most usefull stuff from this communnity. And i feel  that if you want to get into raw you may have better luck looking for processing in astrophotography.

Hope it helps and really looking forward to your findings and/or helping you!

Is mlv dump multithread?
if not, i think there is a way to speed things up a bit.

Just tried converting a few gigs of mlvs with your app tony, but instead of using one window i opened four(as much as the threads in my processor). Seems to work.
So, if possible, you can run multiple instances of mlv dump to speed things up when dealing with multiple files.

Haven't seen anyone talk about it, so i though i'd share. If it's been discussed before, plz point me to it. Thanks

Forgot to mention, i ve been using 1.7.2 with mlv_dump, not mlv2dng.

@arrinki
Well, i haven't see it that pronounced. I usually have to lift the shadows or midtones a little bit for it to appear(at least with h264 and some earlier raw tests).

So in summary: the dark clip must be exactly the same camera, all the same exposure and resolution settings and approximately the same temperature. If any of these change you need to shoot another dark clip.

As i understand, in order to apply the black frame correction, you need to shoot with cap on, but with the same settings(lens, f-stop, shutter, white balance,iso). Is that not the case? Or this is pattern noise visible in the same way, in every combination of the above elements?

Also did you have any pattern noise in highlights or midtones or only shadows(that is without using the raw2dng vertical stripe correction fix)?

So it needs to be applied as a pre-processing step to debayering and couldn't be fixed later(is that certain?).

Is the Line Noise Filter in raw therapy able to fix that? It says it fixes pattern noise.
arrinki if you have some footage with this problem can you test rawtherapee solution as well?

I remember having these lines with h264 video but haven't encountered them using mlv_dump lately.
How much do you have to raise the shadows in order for the stripes to be seen?

maybe off-topic, the only problem with this solution is that you need mlv metadata that you don't have after dng extraction(wb,exposure) or is there a way to obtain them from the dngs?

i guess you don't process the mlv file, rather than the dng sequence.

Just layer them in a program(photoshop, resolve, whatever you are using) and choose a blending mode, subtract possibly the best solution.
If that doesn't work try different blending modes till you are satisfied.

Hope it helps

@ tonybeccar, I have encoutered them a few times with my 7d. Especially when using ETTR, expose to the right and  stuff.
If you raise your lumincance a lot they are easily seen and ugly.

Maybe there could be a checkbox and user can decide if he wants to use mlv_dump or mlv2dng.

@chmee
good to know man! I think your app(with the vertical stripes correction and native mlv support) is the way to go about raw on any workflow(premiere, davicni, you name it) 

Forgot i had an acc here. First post, hi everyone!

I don't think using mlv2dng is good, because i think stuff like vertical band correction and so on is not implemented(maybe i am wrong, but i ve been watching things for some time and i believe
that development on mlv2dng stopped before that). Possibly other things are missing.

Best solution imo would be to use chmee's awesome program, because of the metadata you later get on cdngs(not able yet to have davinci read metadata from dng files from mlv_dump)
Still, even chmee's app haven't implemented vertical stripes yet, so better to use mlv_dump for now(quality-wise)

@Chmee, i thought i read a post(correct me if wrong) where you said that raw/mlv2cdng isn't important anymore coz of premiere and resolve importing dngs natively. I d like to disagree, your
app is very useful, especially if it can tranfer the metadata from mlv correctly(lens info, framerates and stuff).

THANK you both for your time and apps, they are really useful!

wrong thread moderator delete please