VaricamLT: Panasonic VariCam LT

Status
Not open for further replies.
bill totolo said:
I exposed with the in-camera LUT and the exposure tools on my Odyssey 7Q (WF/FC).
I'll look for those files tonight.
Click to expand...

I dont get this.. in terms of noise a small open in Tstop with a consequential darening in post might reap huge rewards?

Exposure is a choice balancing.. easy workflow, highlight retention and noise.

S
 
So on first look I thought the noise was bad, but I dont think so.. when the 800 image is crushed to match the alexa the noise basically goes away.

I tried to match the colours.. (being confused by the actual colour of her jumper and hair) which I found extremely interesting (and hard) to match. I then overlaid a gretag card showing what the actual values should be.

It would seem that both cameras are off but the dark blue has gone a bit nuclear on the vari especially at 5000, which to some extent may be a function of shooting under tungsten and starving the blue channel with subsequent push in camera? A 4k sensor with smaller pixels than the alexa would surely be more vunerable.

It would seem that testing under tungsten may not be the best unless one is doing a tungsten shoot?

https://www.dropbox.com/s/aqseesp2odxqbdz/papvscard_3.5.1.jpg?dl=0

S
 
Last edited:
Morgan,

I just took a look at those files. I'm going to hold off on sharing them publicly because I had a pre-release camera, and that's not fair to Panasonic. I may get a hold of a production model soon, when I do I'll share those files. In the meantime I'll refer to this pseudo test shoot with a production V35:
https://vimeo.com/135406022

The 1st half of this project is noise reduced at ISO 5000. When we get to the wide shot at the corner of Hollywood and Highland we switched to ISO 800 without NR. You can clearly identify the noise on the tile walls of the subway and in the darker areas of the frame. I don't really have a problem with it but as you can see, blacks are crushed, colors are saturated yet there's still plenty of noise at 800.

So here's an interesting thought, if you try to minimize the noise at ISO 5000 on the Panasonic by shooting at say, ISO 4000, that's only a one stop advantage over the Sony FS7. Just food for thought.
 
.. Apart from the 7 is basically 1000 if you treat it the same way.

my basic point is I think exposure levels and post have such huge impact that calling a camera noisy or suchlike is mainly balls. It may have noise in certain situations but most cameras are not noisy per se.

CPs results were so bad that they could impact the rep of the camera. Possibly with justification.
 
Last edited:
Dammit. I feel like a hack, guys. Tied up with so many other things and we shot these two weeks ago. I just realized that these were shot on a V35, not an LT as we couldn't schedule these when the LT was at the rental house. It shouldn't make any difference because the sensor and processing are the same, but I still mis-named everything (since fixed). Between that and burning the 709 into the Amira, it's like amateur hour over here. My only defense is that these were shoehorned into a prep for a commercial and our attention was elsewhere. I'm going to change LT to V35 wherever I can, short of pulling down the clips and re-slating them because I don't have time.

Regarding Sam's comment:
"It would seem that testing under tungsten may not be the best unless one is doing a tungsten shoot?"
Why wouldn't I test under tungsten? I shoot under tungsten all the time. I suppose it's worth testing under both 3200 and 5600 but at this stage of the game, cameras should be able to handle both fine.

Also, Sam--you did grade the Varicam footage from the log clips in the Dropbox, not from the pre-graded footage in Vimeo, right?
 
Last edited:
Not from drop box - missed that link - will retry. Actually what I did is not that relevant - I was more interested in what your person did to make the first video - it seemed to be neither log nor finished but somewhere in between. A status that would really show noise.

Actually I think the burned in alexa seems to present an excellent reference point! It does however favour the alexa if you adhere to the old "bits per stop" argument that says shioting baked on a narrow DR scene presents a better file.

What suprised me was the huge swing In the blue tone and hair tone - the Alexa looke pleasant while the vari not at all - enough that as a producer I would pick the Alex. It then occurred to me that the ct of the lights could impact ESP for the 5000 setting - the vari with the ND on was basically being presented with very few blue photons - that is physics and will not change at any point in the game. For a camera to work under tungsten it basically needs spare ISO to boost the blue channel. Capturing 4k log (14 stops)at 5000 ISO is a hugely different engineering challenge compared to 709hd (7 stops) at 800. Take away two more stops in a channel and the challenge becomes bogglingly hard (I guess)

So nothing wrong with testing under tungsten just it is a factor if the blue goes wonky :)
 
Last edited:
bill totolo said:
I haven't done a side by side, but I've used both. I think the grain on the FS7 is pretty tight at 2000.
Click to expand...

Cant really comment without seeing specifics. I'm gonna present two opposing arguments here..

1) cameras should work best as rated by the manufacturer - therefor one should test at the rated ISO

2) they rarely do so testing without bracketing/ettr rarely reveals the true potential of the camera

I would comment that as one exits the consumer zone where specs shift units and enters the pro zone where specs should be tools in the workflow one would expect a higher correlation between specs and actual

I note with a smile that arri seems to rate all conservatively and reap an excellent image. The fs7 and alexa probably have the same sensitivity.. 800.. But for market sector reasons they say different things in the promotional material and menu.
 
Last edited:
bill totolo said:
Very interesting Noel. I've found the V35 and LT to have noticeable noise at both native ISO's.
Did you overexpose the image a little or simply use the LUT in the VF to expose with (like I did)?

Thanks,
Click to expand...

I have found underexposing too much of the image can give noise. I did the candle test with 3 candles, and found I had very little, so not that much required to get into the safe zone. Ill see if can grab a shot from the camera tomorrow with that and post it here.
 
On the strange jumper (nuclear blue in CPs original 'graded' clip) I went and got the log files from the Dbox link.

gone on the the panny website, DL the panny LUT added it in resolve and voila.. good colour, add a simple Scurve and we are very close to matching the alexa

https://www.dropbox.com/s/b50n8nqmz43wf31/papert_properLut.jpg?dl=0


CP: Im not sure what your post guy was doing it seems he selected some LUT that threw the colours right out of whack (well out of whack compared to the amera and what the colours look like when you add the official panny LUT)
 
It's impossible to guess about the colour without a known starting point. It appears that the posted images have been lightly graded from the original VLog in which case the colour choices are down to the colourist.

EDIT: I've just seen the dropbox links so i'm downloading those, are they source VLog?

The Amira is simply running a 709A variant and the colour choices made are standard Arri - they're good choices.

The comparison would be to take the Vlog and then transform the same colour to 709A but even that isn't totally useful. The whole point of the Varicam is the larger colorspace which is not even possible to compare via vimeo or online.

We're only just moving to the point of wider colour spaces (new iPad pro is DCP apparently) and *then* these cameras will shine, because they can do everything a 709 camera can do but a whole load more.

cheers
Paul
 
Here we can see that in the original 'grade' the blue is well out of whack compared to the 'official' gretag value (RGB:56:61:150).. typical of using no lut or a home-brew lut.

The lower blue is the correct RGB value exactly.

S
papvscard_2.jpg
 
Last edited:
Hey guys, I've been away and have kept up with this thread; has it been announced how much is this cam going to cost?
 
morgan_moore said:
I think we know the intent here.. good clean colour. to some extent that is revealed by the chart

(my first image had the chart actual reference colours overlayed https://www.dropbox.com/s/aqseesp2odxqbdz/papvscard_3.5.1.jpg?dl=0)

You can see the first coloured video the blue is out of whack be both looking at the chart and also seeing that the jumper just looks wrong.

S
Click to expand...

Yes, true. You can eyeball the colours but all these transforms exist because they're quantifiable. So the Panny Vlog->709 is their definition of what 709 footage should look like. But don't forget that it's 709, not sRGB, so the tone curve is slightly different (if you're viewing on an LCD you should be compensating for that too). Also LUTs are not infinite by definition although this LUT looks meaty enough to cover most bases. Creating your own LUTs via LUTcalc means you need to ensure the LUTs are dense enough. An interesting comparison would be to generate a VLog to 709A LUT and compare with panasonic. I feel that will give a closer match to the Amira because the arri 709 LUT isn't just changing primaries, it's doing a complete gamut map - hence the reds being a bit blue on the Amira (which i think it helping skin tones)

There's a varicam IDT for ACES that i will have a play with too.

The difference between the two, if that's important to someone, is that the 709 LUT is clipping out of gamut colours from the camera, it has to. The 709 gamut is much smaller than the varicams. The ACES route should preserve the entire native colour gamut.

I'm a fan of getting footage to a quantifiable baseline and grading from there.

cheers
Paul
 
Using the standard varicam LUT and a simple Scurve the blue feels far less out of whack..
variScurve_3.1.1.jpg

As the difference between 709 and Srgb of course you are correct.. but the arri just felt right.. with the panny and official lut the vari feels pretty right too..
BTW I check colours with various tools.. not visual appearance.
 
Last edited:
I felt comfortable enough with the properly colored Varicam footage (again--none of that is represented in my clips, we looked at it in a color suite but I didn't have it laid down to a drive) that I was ready to move forward with the LT for upcoming show. Unfortunately that isn't going to work out budgetarily so I am reluctantly moving to the F55 (the Canons are out due to no internal ProRes and the less production friendly form factor).

The only real concern I had prior to this with the Varicams was that moire. Would be very concerned about that popping out on a number of other articles of clothing based on what we saw here.
 
paulcurtis said:
YSo the Panny Vlog->709 is their definition of what 709 footage should look like.
Click to expand...
Let me just point out, this part isn't strictly true. Panasonic doesn't (as far as I know) offer a VLOG-709 LUT. They offer a VLOG-V709 LUT. V709 is their stylized interpretation of a better-looking 709 palette. It is not an accurate, strictly-by-the-book 709 palette. V709 is different from Rec.709.
 
Status
Not open for further replies.
Back
Top