music video shot at iso 118 - updated

Started by 70MM13, December 07, 2019, 04:10:19 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

70MM13

this video is pushing the 5d3 right to the limit using iso 118

the setting is similar to the dark music video i made recently for the revenant redux of android rock, only darker still!

the opening shot was done with timbytheriver's new Q-menu build.  it was such a joy to so easily adjust the preamps, plus no waiting for the amps to settle!

regarding the opening shot:  it was by far the most challenging video i've done yet.  there's almost no light and the "LED" readout on the motherboard is unbelievably bright.  it's actually a series of tiny white led point lights, one for each "segment", covered in a red film.  you can see the little points of light and some areas where the film is not entirely covering the spaces.  NO CLIPPING, and the shadows are lifted to the moon.  try that with stock canon iso.  please.

the large monitor on the desk is at its brightest possible setting in a dimly lit room.  there's no direct light at all in the centre of the room.  look at the contents of the monitor, especially while i am playing with the "patch cables" on the virtual modular synth.

as always, no noise reduction at all...

https://www.youtube.com/watch?v=-JDFFq02bBc

timbytheriver

5D3 1.1.3
5D2 2.1.2

Dmytro_ua

5d3 1.2.3 | Canon 16-35 4.0L | Canon 50 1.4 | Canon 100mm 2.8 macro
Ronin-S | Feelworld F6 PLUS

70MM13

yes they are!

i've been using tannoys since 1985!

70MM13

obsoleted update... is there a "delete" function? for obsolete posts like this one?

70MM13

updated again...

i used the new "TNT" build to reshoot the opening scene in even darker conditions!  check it out!

ridiculously dark opening shot!

i will upload a lossless 4k version in the next week or so when i have access to a real internet connection.  sorry about the yout00b compression artifacts.  they are not in the original!

DeafEyeJedi

This is all maddening ridiculously insane! I definitely noticed an improvement in the opening shot. Very Nice!

Looking forward to the Lossless 4K version. Not that we need it. That's how good it is already!  8)
5D3.113 | 5D3.123 | EOSM.203 | 7D.203 | 70D.112 | 100D.101 | EOSM2.* | 50D.109

Luther

That tshirt is the demosaicing nightmare! haha.
Very nice video. Very industrial feeling, as always.

One note: in the 32s scene, the light in the keys are pinkish. But on 1m17s they are red. Don't know if that is inacuracy on the color processing or not. Are you using ACES?

70MM13

@jedi, you make me smile with your wonderful expressiveness as usual!

@luther, your eye is always perfect.  the colour of the leds on the prophet-5 are a big challenge in that light because there is orange light spilling from behind the "tannoy" stained glass window over the door into the vocal booth, changing the colour temperature for that shot very radically.  i am trying to find the right compromise to not have a horrible sickly colour while not having very pink leds!  another issue with the light in that shot is the brightest light from the single overhead light in the room is very VERY cold!  that is why the red leds in my shadow are better...

i am using aces for everything.  i don't even mention it anymore, it's just my permanent setting!  i love it!

i haven't rendered the 4k version yet and i will be doing it tonight...  if you have any suggestions for the tricky prophet-5 shot i would be very thankful!

Luther

Got it! Mixed lights is very tricky, indeed. On Resolve there's a tool called "Hue vs Hue". You can find the right hue of the lights (I think there's a picker) and then shift them to something close to what you see in real life. Sometimes this will cause other tones to shift too (in your case dark oranges, or other tones of red). In this case you can use the red curve in RGB curves to adjust overall red balance or use a mask to isolate only the red leds...

70MM13

yup, i love the ability in resolve to do all the things you just mentioned!

i deplore using masks to fix things, but in this case i may have no choice...  it always feels like "cheating" to me, as though i couldn't fix it the "right way"...

thanks for giving the nudge into my "discomfort" zone! :P

as frustrating as this can be, it's also so much fun!  i just wish there was a way to convince myself that one of these videos was 100% complete!

;)

70MM13

luther,
what do you think of this?

i decided to make no compromises so i warmed up the shot a bit to reduce to cold blue a little bit, and then after isolating the magenta side of the reds (starting from middle red going all the way to blue) in a separate node, i pulled them toward red as much as possible without any side effects.

it's a very subtle change, but i think it is good now!

what do you think?

original resolution image:


jimiz

Quote from: 70MM13 on December 15, 2019, 12:43:42 PM
updated again...

i used the new "TNT" build to reshoot the opening scene in even darker conditions!  check it out!

ridiculously dark opening shot!

i will upload a lossless 4k version in the next week or so when i have access to a real internet connection.  sorry about the yout00b compression artifacts.  they are not in the original!

nice video...where link of "tnt" release?
5D3-123

70MM13

check out tim's thread on cleaner iso presets!

we need more participants, you are welcome to join us!

Danne

How far can these regs be reduced? What happens when "over"doing it? Curious.

70MM13

we're already taking oxfe to 0, which takes us more than halfway there.  that step is "free", as in zero downside.  no loss of dynamic range, only loss of noise ;)

reducing the preamp gain is the step where we can go too far.  the goal is to reduce is only as much as is required to fill the histogram (so to speak) and no further.  if we go further, we start to lose highlights (but with cleaner and cleaner shadows as a result).  my original work with this was going too far, because i didn't really know what i was doing yet, and only following the basic procedure in the first post of the iso research thread.  eventually i started to realize how to make it better, and came up with the "calibration" method, which is now made super easy and more enjoyable/usable by enabling the raw_diags window in the liveview so we can see the instant results of changing settings expressed as dynamic range in EV! 

i show how to do it in the recent tutorial video:

https://www.magiclantern.fm/forum/index.php?topic=24720.msg223628;topicseen#msg223628

try it out!

I know that a1ex said the realtime calculations aren't as accurate as doing graphs, but as you can see from the graphs i included in the thread, it's even better!  the important aspect of the realtime readout is that it is consistent.  all we need to see is how we are doing in relative terms.  when we see the readout reach its maximum value before falling off, we're ready to go!

it would be AWESOME to automate this :)  -- your expertise would be MOST appreciated!


jimiz

Quote from: 70MM13 on December 22, 2019, 01:40:01 PM
check out tim's thread on cleaner iso presets!

we need more participants, you are welcome to join us!

..but you meand the new Q menu system preset (to add on any ML firmware SD card...)   ... OR a special MLraw firmware named "TNT" to wtite on the sd ?!?   :o
thx ;)
5D3-123

70MM13

tim has it on his thread for download, but it is a work in progress.  follow the links on there!