Welcome to our community

Be a part of something great, join today!

Register Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Amira vs Weapon 2K ProRes 444

Kemalettin Sert said:
just WOW.RED color was perfect on Alexa and Amira.Now i see weapon have better RED than Arri sensor.More vivid and vibrant
Click to expand...

To quote myself from February 2014:

Phil Holland said:
Color looks great.
Click to expand...

Dragon's Color Science, Sensor Design, and OLPFs are very big advancements. Weapon seems to make "great" even better.
 
jacob.schwarz said:
So i had a client send me an Amira to shoot on tomorrow because they wanted the ease of a ProRes workflow. I told them that i would rather shoot on the Weapon and I can now deliver them 2K ProRes 4444 just like the Amira right out of the camera. They were totally down, but since the Weapon currently can't play back QuickTime files in Camera for Review we might still be shooting on Amira after all.

But a lot of people even Arri Rave about their color science and sensor, so shot some personal test to see whats up since i had time and both cameras on hand. I'm by NO means Phil but here is a side by side of both color comparison of the Weapon and Arima.


Setting for both Cameras:

Cooke S4 50mm t.5.6
5600k
800 ISO
2K ProRes 4444

Weapon - RedLogFilm / DragonColor 2 (STH OLPF)
Amira - Log-C

I imported both into Davinic Resolve and color matched them to the SpyderChecker chart from Cineon to Rec709.

Here are the results and before i tell you guys which is which i wonder what your best guess are. I'll post the actual ProRes clips tomorrow after my shoot.
MMdx
Click to expand...


What does the percentages under the boxes mean, is it how much each color where corrected to get right?
 
The toe is the giveaway. The colors look very nice. I notice the deep blues on the Dragon are a little over saturated - sometimes we have to drop back to Dragon Color 1 when the blues get too crispy - often with LED lights.

Has anybody posted Dragon/Weapon comparisons yet?
 
Yeah, I've been a Red owner for almost 4 years and you can see the over cooked blues and reds of the Dragon/Weapon. Arri won't match Reds resolution but they definitely have a better color science on their sensor. Dragoncolor 2 also turns the blues up too much for my taste and will clip sometimes.
 
jacob.schwarz said:
I imported both into Davinic Resolve and color matched them to the SpyderChecker chart from Cineon to Rec709.
Click to expand...
Important safety tip: you need to normalize the luminance levels (whites / mids / blacks) before you use Color Match in Resolve. Otherwise, the results are unpredictable and erratic.

Even then, there's nothing Color Match can do that a good colorist can't do, just with scopes and knobs and eyeballs.
 
So you think the % is percentage error, with larger % indicating more correction needed to achieve measured accuracy? So how come the Weapon is measuring more accurate colour? :-)

Graeme
 
I don't know what those percentages mean but if that's how you interpret them, Weapon color is more accurate across the board. (Except lime green, can't win 'em all)
 
Marc Wielage said:
Important safety tip: you need to normalize the luminance levels (whites / mids / blacks) before you use Color Match in Resolve. Otherwise, the results are unpredictable and erratic.

Even then, there's nothing Color Match can do that a good colorist can't do, just with scopes and knobs and eyeballs.
Click to expand...

But Marc, scopes and knobs and eyeballs is not really giving any scientific answer to what camera does the best color representation.

If I understand it correctly the color match function works like follows. I might be wrong and would love to be corrected if thats the case. but my guess is the following:

For both images the white point is taken from the white square for each camera and correction is done to set the white point and balance. Thats why white says 0% for both. Then the other percentages is how far the colors differ from the real values, the difference can be due to a lot of things, the light temperature trough out the spectrum, lens introduced errors etc. But as there is two cameras involved one can see a difference between the two. like Dragon corrects orange by 5% and amira needs a 8% correction for the same color etc. If the values are somewhat uniform (even it´ts progressive from bright to dark or such) then that says it´s a more "even" camera and lens, if they are very odd and different from box to box then things are not uniform and the camera and lens is not really representing the colours in a realistic looking way.

So if I´m right about the above then all comments about one chart looks better than the other is a really odd statement as they have been going trough a process where they been mathematically warped to look the same, no?
 
I personally think Red is more accurate but many people tend to prefer the bit off colors from Arri. More pastels with off blues and reds and a subtle green cast. Like film, it was off too and making images less real. Fortunately we can adjust in resolve.
 
well to be fair whats the settings on both? cuz both red and arri have various picture profiles if you will given the red color gammas and arris X-2 and other profile options...both can have sharpening on and off...it will be interesting to see the specific details of settings since perhaps one can achieve a closer parameter when comparing the two models under other picture profiles? i think its tough to determine whats normals rec 709 since both have a few options up their sleeves is all...
 
Brian Merlen said:
well to be fair whats the settings on both? cuz both red and arri have various picture profiles if you will given the red color gammas and arris X-2 and other profile options...both can have sharpening on and off...it will be interesting to see the specific details of settings since perhaps one can achieve a closer parameter when comparing the two models under other picture profiles? i think its tough to determine whats normals rec 709 since both have a few options up their sleeves is all...
Click to expand...


Sharpening does not do much to flat coloured squares. What ever color profile the adjustment percentages would be even if the camera is balanced / accurate what´s the gamma gain etc is not so so relevant. He states he used redlogfilm and Clog which is both "enough" logarithmic to not clip in any of the cameras in any of the color boxes so inputs should be quite the same, possibly some difference in contrast, gain, gamma and so on but if shot even half right none of that should matter much.

To make the test peaking more into the details one could before doing the color match set the white point, the mid gray point and the black point for each cameras checker board. Then enter the color match tool with both and see how much the different colours deviate from the correct values. Such procedure would eliminate most of the differences in Clog and redlogfilm. and the result would pretty much just show the differences in color accuracy between the two..... I think. I´m not Graeme, I´m guessing and making assumptions based on how I do things when I make color matching for composting.

Graeme Nattress said:
So you think the % is percentage error, with larger % indicating more correction needed to achieve measured accuracy? So how come the Weapon is measuring more accurate colour? :-)

Graeme
Click to expand...

You tell me, you made the camera. :) But yes I guess dragon shows smaller numbers of deviation which is, if I´m not mistaken, proof of a more accurate color science. I don't use resolve so I do not know what those % numbers are. Possibly somebody here that do use resolve can confirm or give a hint of what we are looking at.
 
Tom Van said:
I don't know what those percentages mean but if that's how you interpret them, Weapon color is more accurate across the board. (Except lime green, can't win 'em all)
Click to expand...
Why would you think that the amount of percentage change determines whether or not one camera is more "accurate" than the other? You are taking a log image and moving it into rec709 color space. The Alexa sensor is well known to create a very "loggy" image (very desaturated colors and very low contrast) and is also well known as reproducing colors very accurately, especially in mixed lighting situations. What difference does it make how "much" the log image has to change as long as the end result is accurate. By your reasoning shooting the chart in rec709 mode (which you can do on the Amira) would be the "best" and most "accurate" way to get colors as the change would be very little.
 
Mark Wuerthner said:
Why would you think that the amount of percentage change determines whether or not one camera is more "accurate" than the other? You are taking a log image and moving it into rec709 color space. The Alexa sensor is well known to create a very "loggy" image (very desaturated colors and very low contrast) and is also well known as reproducing colors very accurately, especially in mixed lighting situations. What difference does it make how "much" the log image has to change as long as the end result is accurate. By your reasoning shooting the chart in rec709 mode (which you can do on the Amira) would be the "best" and most "accurate" way to get colors as the change would be very little.
Click to expand...


Yes of course if you took one of the images and brought down the contrast to half before doing the colormatch the numbers would increase. So if those are numbers of deviation then higher numbers does not really tell anything. but if blue got 50% and a color next to it got 5% or such then one can assume that such sensor, or input image, is not really accurate and would need a pretty warped 3D mesh in the 3d lut to become correct. The more twisted the mesh the more deviation between the matching points, where as an image that gives even numbers / having more of a uniformed shaped warp mesh are more likely to be correct. or?
 
You must log in or register to reply here.
Back
Top