seku

this is what we already hoped for with the initial Scarlet (prosumerish device). Didn't pan out that way unfortunately ... but count me in if this happens :D

hi @Sage!

just bought your 2 LUTs, looking forward to working with them

nice! you're getting the hang it!

if you feel like taking the next step : try replicating that same look in davinci resolve.

back when i tried to "get" davinci, i did a conversion from a DNG in capture one pro i really liked. then i imported that image into davinci resolve. and took hours and hours to get it to look the same.

TLDR: it is possible to replicate looks from ACR in Resolve. but it is quite the learning process (hint : Arri LogC is a lifesaver)

i hope i am not too harsh, but there is no 4k ML raw. closest you can get is 3.5k wide, but that's considerably more squeezed than 21:9. Also, you'll get a few seconds at that resolution at best. Also, consider that ML RAW has no highlight rolloff. it clips quite harshly. because it's raw the highlight rolloff you saw in the videos was probably handed quite handsomly by a gifted colorist who recorded ML RAW with enough leeway to let it roll off nicely in post.

Flexibility in grading? oh yes. ML RAW is awesome for that. but you need to learn to shoot with grading in mind.

For your bonus question: the 5Dmk3 is meh for DR. chroma noise creeps up quickly. you don't need any white balance, as this is RAW. i personally would not go above ISO 3200 with the 5dmk3.

Hi all,

just got my Rectilux hardcoreDNA, and was wondering what you guys would recommend as diopters. As the Rectilux focusses up to 70cm, diopters don't need to be too strong? like something from 0.4 to 1?

Thread size is 86mm on the rectilux, so choice seems limited. There seem to be some very cheap 20-ish bucks Vivitars on ebay. Also, Heliopan seems to make 86mm ones as well, but they seem exceedingly rare.

Any thoughts?

your prized Kowa B&H mods? i really wanted to ask you about those. really glad that you're working on them. i'm quite interested in those.

(just one hijacking question : how far do your mods take the desqueeze? from 2x at infinity to how much "compression loss" at near focus? 1.5x? more, less?)

@liamlumiere : doubtfully, i orderedmine at around december  

i personally prefer neatvideo. by quite a margin. but it is easy to overdo stuff because it is so capable

From what i remember from the Leeming Lut creator and @Sage, both state that dynamic range in Vlog-L mode and HLG is roughly similar.

The difference is that V-log-L is made to intercut with Vlog from the EVA, and uses a rather shallow portion of the available bit range, whereas the HLG gets a more full-swing approach to the available range.

Theoretically speaking, there should be more tonal definition in HLG. 

My main question is tho : how to expose HLG?

@liamlumiere : it arrived today, can't wait to get home and play around with it  I ordered it via his facebook page, John was very fast and courteous in responding

so this thing does pixel shift 4times it size AND is single-chip DLP (which means it needs to expose each color seperately).

so if we take 60hz input, we would need at least 4x3x60hz = 720hz on the DLP chip? 

Fast little bugger...

About supper choppy livemode : i think @Django thinks of crop mode, where monitoring sure is clunky. (1 image per second / gritty grayscale). Also framedrops after a few seconds because you exceed the capabilities of the CF controller. That's the price for being on the bleeding edge.

Then again, i filmed a whole 3 week vacation with 3k 21:9 RAW, 7seconds maximum in 10bit (compressed wasn't out back then). with sound. I still need to work through those 2 teras of footage...  

yep, MLVApp works great as well.

 @Django : agreed on hit&miss. sound works fine, unless you use compressed RAW mode. (and i think they added mlv_sound to mlv compressed crop just a few days ago)

if you use Resolve the workflow is very easy with MLVFS tho (it takes me less than a minute to ingest some 100 clips into Resolve with sound).

actually... i am cameraless right now. (the screenshot above is 5dmk3 RAW in fullframe binned mode, not even cropped)

(as much as i love the RAW of the 5D ... i want a waveform for monitoring. When i got exposure right, i love the 5D. The histogram helps me with ETTRing, and making sure i don't overexpose. But i have no clue where my skintones fall).

Always wanted to try out anamorphic though. So i think tomorrow i'll buy a gh5 & GHalex LUT to go with my schneider cinelux 2x and rectilux. But that is quite a bit off-topic

No cinema cam for me tho, i'm just a bloody hobbyist

just for fun, here's a screenshot from one autumn japan trip, edited in resolve with the workflow i described above

@AZGeorge: you are just getting static noise?

Did you by any chance use one of the nightlies that enables lossless compression? Then you also need to use the compatible dump or mlvfs builds that supports compressed crop mode

just a short explanation: crop mode ... just takes a crop of your 5d sensor (quite close to m43 actually) but uses all pixels in that rectangle. Meaning the OLPF works perfectly, and there are no resizing artifacts whatsoever. i filmed with it last year, and it was rather experimental. when it did work, it was fabulous.

@kidzrevil : yeah, i think Resolve really is worth the "trouble". I recommend googling Juan Melara for getting some grading ideas and to get your mind blown.

btw, if you don't like binning, be sure to try out crop mode. it's the reason why i bought my 16-35mm back in the day. no more binning, just damn sharp 2.5k ... or even 3k 21:9 if you are pushing it. Beware tho that at those resolutions, the buffer might fill up after 6-7 seconds. The footage is gorgeous tho.

nice looking stuff

i'm no longer shooting on 5d RAW, but my workflow was rather easy:

1. Mount the MLV files in MLVFS (this mounts a new virtual hard drive where all of your files miraculously transferm into cinemadng in real time)

2. Import the files from the virtual drive into Davinci Resolve. Transform the RAW into linear, then linear to Arri LogC.

3. Cut and grade at leisure.

4. Export as dxnhd or your preferred codec

Well... i was interested in how well the GH5s compares to the GH5, 5dmk3 and the a7r3 in terms of lifted shadows on RAW images... just to see how good the sensors are.

From the recent dpreview, here's a screenshot of the 4 cams at +6 stops ... 

Imho, the raised shadows from the GH5s and the GH5 seem similar. They both beat my old 5dmk3 handily.

But damn, that A7r3 ... that image quality should be forbidden. If only its codec could keep up

Rectilux is doing its thing. waiting for my hardcore DNA 

from @Andrew Reid from the main article:

Quote

I will be testing these extensively and finding out which to choose for my next version of Pro LOG, taking the guess-work out for you. I’ll do the same for the upcoming 7S III which will be the first Sony Alpha to shoot 10bit.

A7S3 is confirmed to be 10bit?

now that would make choice even tougher than it already is.

I couldn't believe 12+ stops either in VLOG-L, but Samuel Bilodeau  from Mystery Box (they did the official "Horizons" Panasonic HDR Short for the GH5s) states that you can use 14 stops under nearly all circumstances, with detail up to 16 stops (very noisy tho, but detail still there).

The TLDR juicy stuff according to Samuel Bilodeau  :

Quote

After throwing up the footage we shot on our scopes, we're measuring the dynamic range captured by V-Log L at around 16 stops - after converting to HDR10 (PQ Curve) our brightness distribution runs between 0.01 nits on the low and 1015 nits on the high clip, for a contrast ratio of 10,150:1, or 16.63 stops. The lower 3-4 stops retained are increasingly noisy, but the detail is there. 14 stops are fully useable under nearly all circumstances, according to what we're seeing.

And the explanation in all its glory :

Quote

I'm not an engineer at Panasonic. All I can tell you is what we were told, and about what I'm seeing on our scopes when we record video as I mentioned in a previous comment (V-Log L preserving around 16 stops of digital data).

I do understand sensor design, however, and I will add an important clarification to what you hear in the video - a 14 bit sensor does not mean that it offers 14 stops of DR, the man says that himself. Dropping the effective bit depth at the ADC to 12 bit doesn't mean that it's lost two stops of dynamic range, you're just sampling each pixel with two fewer bits per sample - 4096 levels / sample instead of 16,384.

You're operating those samples on the same dynamic range of the linear light as the 14 bits, but you're decreasing the precision of detail within each part of the brightness range evenly, increasing the SNR without decreasing the DR. And that's assuming you're operating your ADC in a linear conversion mode - ADCs may also run in a log format. Again, I'm not a designer of the camera and I don't know if V-Log L runs on the ADC or as part of the color processing engine; I assume that the ADC operates in a non-linear mode because most ADCs do to preserve more detail across all stops of light.

Keep in mind that the final V-Log L output does not run the full set of available digital values on a 10 bit log scale, meaning it compresses further each unit of dynamic range (each stop of light) so that each is more-or-less allocated an equivalent number of bits for detail retention.

Whether you record with V-Log L on camera or using an external recorder, the images captured are equivalent in this regard.

Also juicy:

Quote

Oh gotcha, and yes - we only used the external recorder for ProRes recording of the 60p content. We did a comparison between the internal 10b 4:2:2 all-Intra H.264 and the ProRes 422 HQ and there's no difference in color correction, dynamic range or latitude that we could tell. You'll need the high capacity high speed SD cards I mention at the top of the article, though. We had a great experience with Panasonic's V90 SD Cards.

This makes me all giddy.

Btw, I love their blog, incredibly useful HDR explanations and workflows.

thanks a lot!