Messages

Thank you all for your very good advices!

Sounds like a lot of work, especially when there is lots of out of focus stuff in the foreground.

Well, perhaps, instead of masking only the lady (a lot of work indeed) I could mask the entire "corridor" of her walk (...perhaps...) and in this way avoiding all the transparency issues of the out-of-focus grass and stuff in front of me... We will see...

What software are you using for post process, or what is your workflow ?

Switch (for MLV to cDNG) + Davinci. That's it.

Another thing or 'trick' to consider, cut and lift the shadows, so all noisy dark parts of the frame become a noise free grey color.

Damn good!!! Thanks a lot for this trick!

i say buy a pair of cheap LED construction lights, add diffuse filters and gain +2 stops of light for your scene. Also buy some tiny LED flashlights and place them behind objects in the scene with some diffuse on and you can "fake" more light in the scene.

Eh... Not bad indeed! I'll try surely!

i believe that 3x3 readout has a cleaner output because of the interpolation, but slightly unsharper.

Then... I'm an idiot (damn me): all my shots are 5:3 (1920 X 1150) ONLY to have some reframing room in post... So, if I correctly understand, 1920X1080 gives me less noise thanks to the interpolation? (Honestly I tried this evening to shoot a semi-dark room both in 16:9 and 5:3, and the result was actually pretty similar...)

And stick to fps override and shutter fine tuning, always.

My English is very poor... "Stick to" means "don't touch"? (Again, I must be an idiot...) I've always keep FPS override activated, for all my shots... But always keeping the same fps ("exact") value of the Canon menu (25). Is it possible that I've made my shots worse somehow?

Edit: also look in to mlv app's Darkframe subtraction, it does wonders when you go >1600 iso

Never used MLV App and I don't know what is darkframe subtraction... But Inoticed "darkframe average automation" in Switch.
Q: Is it the same thing? How does it work?

The color profile tells the raw what the range of color values to use and the noise is colors that are out of gamut

So, if I correctly understand, I could anyway take my shots in RAW and then converting them in Prores only when I find  some out-of-gamut artifacts (like too much noise). Or this operation must be done in-camera?

Again, thank you all a lot!

Thanks a lot for your reply Levas.

Also wondering, what noise are we talking about, is it mostly the grainy noise, or do you have vertical lines/banding noise, or do you have static, bright coloured pixels across the frame ?

Mainly grainy noise, but I noticed also vertical lines!

So I would definitely recommend your own suggestion of faking the shot, start shooting as soon the street lamps are on.

Yes, that was my first solution. But what about the second? (Compositing of sidewalk-timelapse + woman-normally-walking?) In this way only woman will be noisy. But what worries me is the fact that camera position is near ground and I see woman "through" a lot of out-of-focus grass and stuff, so I don't know if I will be able to make a realistic comp in post...

P.S. What about Override in your opinion? Any advantage in keeping it active at the same exact FPS (25) of Canon menu?

There is no difference in noise between shooting 16x9 or 3x2. You are shooting raw, the ratio only determines how much of the sensor raw data you want.

Is this true also between "normal" 16:9 and crop-zoom 5X? I mean, I can suppose that in 5X crop mode the amount of light is reduced, then I should have more noise... or not?

Hi everyone,

is there any benefit (or disadvantage) in keeping FPS Override activated even when it's set at the same frame rate of the Canon menu?

For example: in Canon menu I have 25 FPS, and also FPS Override is at 25 FPS (exact). In this case, is there any benefit/disadvantage in keeping FPS Override activated?

Thanks

[...] but I am of the unpopular opinion that noise is inherit in the cDNG workflow.

Possible? Is there a technical reason?

The above numbers apply to 5D3 in 1080p mode (3x3 binning). Things will be different in 1:1 crop mode (ISO 6400 no longer providing a noticeable benefit, but that's not applicable here), or on other cameras (where the DR figures may be slightly different).

This could be the point! I'll surely test 6400 ISO in 16:9 1080p (perhaps the problem was the 5:3 ratio I used mainly to have room of re-framing in post).

[...] with FPS override [...]

Very-newbie question: even if I shoot always in 25fps, in ML menu I usually keep FPS override always activated (both Canon menu and ML override are set on 25fps). Is that a mistake? Should I deactivate ML override when shooting at 25fps?

Thank you all for your replies.

feel free to use ISO 6400.

Very interesting topic! I didn't know it. You are the wizard here, Alex. Anyway – please forgive my skepticism – even just at 3200 ISO I see a lot of noise (noticeable worsening than 1600), so I'm afraid of using directly 6400 ISO...
Perhaps my issue could be related to the ratio I use? I usually shoot at 5:3 (1920x1152). Do you think that shooting with 16:9 ratio could improve S/Noise performance?

Photo or video? Are you shooting raw? What's your post processing workflow?

Thanks for your reply Kirby.

Video: more precisely, my first solution is thought as video only, while the second is a combination of video (walking lady) over the timelapse (photos).

Uncompressed RAW 14bit.

My workflow is MLV-to-cDNG through Switch, then directly into Davinci. Nothing more.

Hi everybody,
newbie question here: I'm going to take a night shot with 5D3 + ML (the scene is a lady walking on a sidewalk, the camera is moveless on tripod, the only lights are a street lamp and the moon).
I made a test with Sigma 35mm at f/1.4, T/30 and ISO 3200. The result was a very noisy shot because the high ISO, but even setting ISO 1600 doesn't help because image is darker so noise appears anyway.
I'm a bit frustrated...

I've thought a couple of solutions:

1. I could try to shoot not-yet-in-night, when the sky just start to be darker but is not black, and then lowering brightness in post to obtain "night effect". But the truth is that all the shadows created by the street lamp are visible only when sky is "noisily" black;

2. I could take a shot of the walking lady when the sky is not-so-black, and later I could take a "clean" timelapse of the street during night, and then masking the walking lady in post in order to add her over the "clean" street in timelapse. Interesting effect perhaps, but again the lady will not have consisten lights and shadows on her in coparison with the night street.

So, how could I solve it?

Thanks a lot

Thanks Squig and Deadcode for your replies.

After a while I decide to keep my 5D3s instead of switching to BMPCC4K. I'll share my thoughts here, hoping it could be useful for someone else in my same conditions (and to clarify myself too).
Giving that low-light and noise performances are roughly equivalent on both cameras, for my way of working (and for the fact that I'm a generic artist, not a specific pro) I see 3 PROs and 3 CONs in switching from 5D3 to BMPCC4K:

PROs:

1. Weight and size: BMPCC4K is smaller and lighter than 5D3, its 5" screen can (often) replace an external monitor, and very-fast mFT lenses (that give same speed performance than SpeedBooster+FF-lenses) are compact and light as well. That means having at least 1 Kg less in your hands, and 2-to-3 Kg less in your bag (and this is not secondary for who has a good arthrosis

2. B-RAW: honestly MLV/CDNG workflow is a little nightmare: time consuming, storage consuming (with the price of my second RAID I could buy a complete set based on BMPCC4K!) B-RAW means having a quality-effective 3:1 file that requires 1/3 of the storage, can be directly drag into Davinci and is fast;

3. Dynamic Range: obviously improved over the 5D3 (never been a DR champion in comparison with other DSLRs)

CONs:

1. Selling 2 cameras to buy only one: in order to take the complete advantages of mFT set compactness and lightness – so, to buy BMPCC4K + mFT lens-set – I should sell both my two 5D3s and all my FF lenses. That means working with one single camera instead of two! And I must say that working with 2 cameras is a great thing in various situations;

2. 5D3 is a very good still camera: that means huge timelapse quality! And as I wrote above, being a generic artist means that my working "paradigm" depends on each different project and can strongly change from time to time. So having a camera that can also take good stills is a big advantage;

3. MagicLantern controls: it's not secondary that ML gives you a bunch of very detailed controls (and functions: see crop mode...) that lack on BMPCC4K. Obviously you can do the job anyway, but things like the spirit level or the customizable cropmarks helped my work really a lot. In order to minimize this lacks on BMPCC4K, I could buy a full featured external monitor, but it starts to go in the opposite direction than "cutting ballast"...

FINAL THOUGHTS:

All 3 PROs represent improvements, that means BMPCC4K and 5D3 do the same things but BMPCC4K can do them a bit better/easy.
Instead all 3 CONs are lacks! That means 5D3 can do things that BMPCC4K can't do! THIS is the answer. That's why I remain with my 5D3s.

Eventually, I could opt for a compromise: selling only one 5D3 (and adding 1K€ more...) in order to buy the BMPCC4K body-only (no mFT lenses) + SpeedBooster. In this way I could keep one 5D3 + one BMPCC4K and sharing le FF lenses between the two. But I remember that I bought my second 5D3 (with many sacrifices) just because it was discontinued, and once my first camera broke down, I would have lost the best/only camera for ML. So, again, for the moment I'll keep my 5D3s

With a speed booster, the BMPCC 4K is just as capable as the 5D in low light.

Q: shallow DoF. I use 5D3 mainly with f/1.4 primes to obtain a good shallow DoF (I usually don't use wide lenses, so in my case, I'd use Speedbooster only for increasing the "speed").
Now, let say we have two scenario: 1. BMPCC4K + SpeedBooster + f/1.4 EF lens, 2. BMPCC4K + f/0.95 mFT lens. Which of these two options can go closer to the percentage of in-focus/out-of-focus of 5D3 + f/1.4 EF lens? Are the two options equally close?

I've shot on the 5D3 before and as I recall it got pretty noisy at 1600 and fell apart at 3200. The BMPCC is clean at very high ISOs.

That's exactly my point: I'm sure that 5D3's FF sensor captures more light than a mFT sensor, that's obvious, BUT 5D3' sensor is less noisy only at low ISO (ideally 100, max 800), while BMPCC4K has native 400+3200 ISO that should mean that even if I need to push up the ISO to capture the same amount of light of 5D3, I should anyway obtain a less noisy image than 5D3. For example: let say that to obtain two images with the same amount of light, 5D3 is set at 1600 ISO, while BMPCC4K is set at 3200 ISO or even more, and the result should theoretically be that BMPCC4K captures less noise.

Could it be?

micro four third a big no for me.i am a owner of both 5d mark iii and bmpcc(not 4k) micro four third very bad in low light [...] 5D III [...] its a full frame much better in low light, better battery life

Thanks Andy for your reply. That's fundamental for me: if BMPCC4K can't beat 5D3 in low-light/noise (in spite of BM native 400+3200 ISO) or at least equivalent, then I remain with 5D3!

Don't forget about Silent Pic Mode for timelapses. (Stills for video) 

Eh... That's a point for 5D3!

I'm not sure about your questions as I don't have BMPCC4K.
But the main difference is that you get a great professional camera (5d3) not only for video, but also for stills.

Thanks Dmytro for your reply. Yes, of course 5D3 is a great still camera (anyway, I noticed that I take very-very few still in my work).
So mainly I'm interested into a comparison between 5D3+ML vs BMPCC4K in video-only territory.

Hi guys, also I have a little doubt about switching from 5D3 to BMPCC4K, and also I'm searching for answers...
I bought 5D3 ONLY because I could mod it with MagicLantern! Because my FIRST need was working in RAW, and the second main need was working in low-light condition without noise (and full frame sensor helps to maintain low ISO and low noise).

Now, I's true that BMPCC4K has a tiny sensor, but it seems that its native ISOs are 400 ISO (and 3200 in Dual-ISO). So, if I'm right, this should mean that BMPCC4K at 400 ISO should take at least the same amount of light of the 5D3 at 100 ISO, without taking more noise!
Q: Do you think it's right, or not?

Second: 14bit Unconpressed RAW is fantastic but 5D3's dynamic range is narrow, while 12bit BM RAW 3:1 is better used by the superior dynamic range of BMPCC4K.
Q: Do you agree?

Third: shallow DoF... Obviously 5D3 + a good F/1.4 lens can reach easily a shallow DoF, but you should reach a very similar result with BMPCC4K + Voigtländer F/0.95 lenses!
Q: Possible?

If the answer to all 3 questions is YES, then there are NO reasons – to me – not to switch to BMPCC4K: lighter RAW, faster workflow, lighter/smaller camera and accessories, better dynamic range, no need for external monitor (5" on-board!)

Is there any other advantage of having full frame?

Thanks

Thanks Levas for the reply.

Well, please correct me if I'm wrong: with Dual ISO we have one ISO for bright areas (usually a low ISO) mixed to a second ISO for dark areas (usually an higher ISO). This sounds to me like a sort of brightness-equalization of different areas of an image. Dual ISO makes this sort of equalization gaining the sensor for the secondary higher ISO. Instead (in my example with an ambient light in the room) I'm trying to do the same equalization simply gaining the ambient light itself (making dark areas less dark...)
Is that all wrong?

Hi IDA_ML, thanks for your reply.
Yes, DUAL-ISO is an option, and it's surely an improvement in dynamic range, but... 720p...
Anyway, don't you think that the soft light in the room should work like DUAL-ISO? (It should work like a sort of light equalizer, no?)

Whitelight, thanks for your reply.
Anyway, the camera is not moving, but the woman interacts with many objects in the room. I can't do green screen here.

Hi everyone,

I'm trying to make a sort of low-key video shot: in a dark room there's a woman watching at the window that should be the only light source.
If I make the shot as I described, directly with the "final" low-brightness I desire, the histogram is very low, and my shot is full of noise.
I tried to raise up the ISO, but the few highlights (the woman's face) starts clipping.
A solution could be using a soft light in the room in order to reduce the difference between the highlights and the shadows. In this way my shot should be less noisy, but it will loose the "final" low-brightness I want. Anyway I should be able to correct this in PP (lowering the brightness-under-a-certain-value to "bring back" the shadows to their darkness).

Do you think this is a good method, or it's better shooting in the desired "final" light + using a de-noise in PP?

Thanks in advance.

Sorry Danne, I've deleted the original MLV after converted into DNGs, and now I have only the corrected DNGs.
But in the next days I will have to repeat the same shot, so if the issue will happen again, I'll post to you the MLV.

Thanks a lot.

I found the solution in ExifTool Forum:

basically I have to use ExifTool in Terminal:

1. knowing the black and white levels of the first DNG frame

Code Select Expand

exiftool -G1 -blacklevel image.dng
and

Code Select Expand

exiftool -G1 -whitelevel image.dng

2. setting those values for all the DNG frames of the folder

Code Select Expand

exiftool -blacklevel=xxxx -whitelevel=xxxx -ext dng -r YourDngFolder

This method works: now I have no more strobe efx.

Hi everyone,

sorry in advance if this question has yet a well known solution. My problem is that I have Dual-ISO footage (5D3 with ML) and I converted my MLV files with Switch.
I obtained a DNG video with a visible strobo effect (in Davinci).

In THIS page there is a video that shows the solution: fixing the black point with ExiftoolGUI, but this software is for Windows only. I use Mac, which is the procedure to fix this problem in Mac? I tried pyExifToolGUI, but without success. Same thing for ExifToolGUI via Wine.

One solution could be using Exiftool from the terminal, and knowing which is the command line for finding the black level of the first DNG image, and the command line to set that level for all the other DNGs of the folder.

Please, someone can help me?

Thanks a lot.

Thanks so much

Hi everyone,

I've just noticed a strange black little soft-spot on my recent shots (see attached image). I say "strange" because it's always on the same position in the frame, but NOT on every shots!
Naturally I thought it was same dirt on one lens, but it's not, because it's visible on shots made with different lenses.
So I thought it was dirt on sensor, instead, but - as I said - it is visible NOT on every shots.
The only one solution I can give (but I'm NOT an expert) is that it's a little grain of something (dirt, ash...) in short something "three-dimensional", and sometimes, when light has a peculiar angle, I can see its shadow...

Just 2 questions:
1. What is it in your opinion?
2. How can I solve? (Service?)

Thanks a lot.