- Banned
- #41
Mike Nagel
Banned
how is the reference source developed in all of these ? IPP2 ?
Welcome to our community
Be a part of something great, join today!
Improving the Camera: The Image
- Thread starter Hrvoje Simic
- Start date
- Thread starter
- Banned
- #42
Hrvoje Simic
Banned
Further improved greenery and re-posted in last three foliage examples.
- Thread starter
- Banned
- #43
Hrvoje Simic
Banned
- Thread starter
- Banned
- #44
Hrvoje Simic
Banned
Fly butter, fly.
Aaron Lochert
Well-known member
All of these stills so far are pre-firmware fix, correct?
- Banned
- #46
Mike Nagel
Banned
it looks good, but I wanna see a direct comparison with the 1D matched (at least in the 25-75 range) as close as possible... this is what is the most important. Once u deviate off the reference 1D, color will inherently look different so it's almost impossible to judge other than the blatant obvious that the artificial-ness of some colors has been muted and finer gradation detail is present (which is a 1D adjustment)...
- Thread starter
- Banned
- #47
Hrvoje Simic
Banned
The second image is not attempting to match the first but improve the image response based on signal analysis and other things used as a reference, which makes some of the image aspects go into different direction. These are only examples of some of the outcomes, without additional grading. If matching images seems as useful, you can do any desired matching yourself when this is publicly available, my priorities differ. 
- Banned
- #48
Mike Nagel
Banned
clearly it's not trying to match, as u're creatively adjusting the image... subjectively to a more pleasant starting point ;-)
but - just to re-iterate my point - it would be helpful to see just color improvements (ergo 3D adjustments off the main axis), without touching the 1D... there are a few reasons why that is important in direct A/B comparisons...
I'll give u one reason, quick scoop: u replied that this could potentially save colorists time. u're clearly throwing off the 1D, which now needs to be set again. so, we lose time.
or does this treatment come into play before the image gets normalized ?
NOT trying to throw this under the bus - quite the opposite - but again apples to apples would be nice and/or ur recommendation where it sits in the processing chain...
but - just to re-iterate my point - it would be helpful to see just color improvements (ergo 3D adjustments off the main axis), without touching the 1D... there are a few reasons why that is important in direct A/B comparisons...
I'll give u one reason, quick scoop: u replied that this could potentially save colorists time. u're clearly throwing off the 1D, which now needs to be set again. so, we lose time.
or does this treatment come into play before the image gets normalized ?
NOT trying to throw this under the bus - quite the opposite - but again apples to apples would be nice and/or ur recommendation where it sits in the processing chain...
- Thread starter
- Banned
- #49
Hrvoje Simic
Banned
I see another attempt of guesswork as another poster in the start of the thread, although I wrote something else a few times. I'm going to assume it is well meant, so I'll try one more time.
No Mike, color accuracy is not being designed by "subjectively more pleasing" premise but signal analysis with regards to established reference. Focusing on factors some would term as "scene referred" with IDTs.
Only a smaller part of signal mapping is subjective, a part which has to be, and that part also uses as a basis many points of reference and care is taken for getting the most natural result.
And as I said you'll be able to do that as much as you please and do direct comparisons how ever you prefer if that is your priority, when this is publicly available. I am posting direct untouched images exported with standard settings and solution applied on new gamma and gamut mapping.
In the case of flying butter with wings I only compensated luminance distribution just to get images more closer, because default choice from RCX is more boosted and has losses in higher signal range.
You are using guesswork as a basis for conclusions. I'm not throwing off anything. Both images are exported with in camera WB metadata with colors untouched, it only seems that way because the signal deviates in the first examples, whereas in the latter it is corrected by the solution.
You won't loose time.
Save your judgements for direct experience, those who have it don't seem to have any problems.
- Banned
- #50
Mike Nagel
Banned
Hrvoje,
for somebody who is splitting hairs all the time and so anal about terminology, u are off ur game today, buddy... ;-)
just saying friendly, nicely... chat between forum colleagues
(a) of course we are all guessing here: because u are not answering questions or sharing direct information. Yet. Which is fine, u're still working on it.
(b) u need to look up "color accuracy". that is not what u are doing. U are doing color adjustments (based off bayer sensor data that is already, inherently flawed and inaccurate). Absolutely fine (we are all here in this thread for creative purposes), but there is a very much a difference.
(c) if the first image is irrelevant, then why are u posting it ? ;-))) of course it is not irrelevant because u a re doing a comparison between solutions.... which needs a reference.
(d) I could go now into the importance of the 1D... which I touched on now 2 or 3 times already... but clearly it is not clicking or simply understanding is missing. it doesn't matter, I was trying to help out to put the solution better into perspective.
I'll try one more time, from a colorists perspective: I have a fixed 1D for a specific shot, which I don't want to change. let's say it is the one from ur ref image. I wanna see ur solution match the 1D (at least in 25-75) and then compare ur color improvements. Ergo, isolating just the color adjustments showcasing possible improvements between different solutions.
that is the whole point of a comparison when posting 2 images.
My 2 cents. But I digress.... back into... (wait for it).... (waaaaaiiiitttt).... (wait even longer).... THE DOMAIN ABOVE THOUGHT !!!
P.S.: Btw, just to put some standard into this thread: the bearded man in a barrel was a reference to Diogenes.
for somebody who is splitting hairs all the time and so anal about terminology, u are off ur game today, buddy... ;-)
just saying friendly, nicely... chat between forum colleagues
(a) of course we are all guessing here: because u are not answering questions or sharing direct information. Yet. Which is fine, u're still working on it.
(b) u need to look up "color accuracy". that is not what u are doing. U are doing color adjustments (based off bayer sensor data that is already, inherently flawed and inaccurate). Absolutely fine (we are all here in this thread for creative purposes), but there is a very much a difference.
(c) if the first image is irrelevant, then why are u posting it ? ;-))) of course it is not irrelevant because u a re doing a comparison between solutions.... which needs a reference.
(d) I could go now into the importance of the 1D... which I touched on now 2 or 3 times already... but clearly it is not clicking or simply understanding is missing. it doesn't matter, I was trying to help out to put the solution better into perspective.
I'll try one more time, from a colorists perspective: I have a fixed 1D for a specific shot, which I don't want to change. let's say it is the one from ur ref image. I wanna see ur solution match the 1D (at least in 25-75) and then compare ur color improvements. Ergo, isolating just the color adjustments showcasing possible improvements between different solutions.
that is the whole point of a comparison when posting 2 images.
My 2 cents. But I digress.... back into... (wait for it).... (waaaaaiiiitttt).... (wait even longer).... THE DOMAIN ABOVE THOUGHT !!!
P.S.: Btw, just to put some standard into this thread: the bearded man in a barrel was a reference to Diogenes.
Anton Esteban
Well-known member
- Joined
- Aug 9, 2014
- Messages
- 116
- Reaction score
- 0
- Points
- 16
Where are all the skintone examples?
- Thread starter
- Banned
- #52
Hrvoje Simic
Banned
Hrvoje,
Looking are the before-after shots in this thread, it seems to me that :
- Yellows and oranges get more saturation, and less brightness
- Greens get less brightness and lean more towards cyan
- Blues get less brightness and lean more towards cyan
All in all, looks very nice.
One problem, though - Whites get a cyan color-cast.
Looking are the before-after shots in this thread, it seems to me that :
- Yellows and oranges get more saturation, and less brightness
- Greens get less brightness and lean more towards cyan
- Blues get less brightness and lean more towards cyan
All in all, looks very nice.
One problem, though - Whites get a cyan color-cast.
- Thread starter
- Banned
- #54
Hrvoje Simic
Banned
Hey Eric, thank you for your impressions.
The signal of every camera deviates on micro levels across its range. In these examples at first you may pick up differences through the reference of primary and secondary hues and the previous image but if you look more closely you may notice that each main hue gets more separation, inner tonalities and more hues pop out. It is much more complex. In the example of skies there are more bluish hues visible instead of one global cast, in the case of greens you may see more different green hues emerging giving overall richness of the palette. Especially visible in the house image example. What is done here is done on micro levels of the signal.
Regarding the cast - this solution is greyscale neutral. What it also does is make tonalities visible which typically digital cameras "choke". Such as those close to white. There is no pure white in object reflections in Nature, there is always some very subtle cast close to it, depending on light properties and ambience reflection, which the observer's mind compensates. In the house example you may notice the warm Sun cast on wooden surface and on the bark of the tree, which is not visible in the first example but makes the image come alive. Taking care of those micro levels is very tricky and takes time and Helium treatment is still in the works.
Also, these solutions are designed to take care of the fixed offsets of the signal by measured reference behavior of the camera, not to compensate for unknown variables of light, that part should be done manually in post, and once the signal is treated that becomes much easier. Touch of white balance compensation and things like these tend to sink in.
The signal of every camera deviates on micro levels across its range. In these examples at first you may pick up differences through the reference of primary and secondary hues and the previous image but if you look more closely you may notice that each main hue gets more separation, inner tonalities and more hues pop out. It is much more complex. In the example of skies there are more bluish hues visible instead of one global cast, in the case of greens you may see more different green hues emerging giving overall richness of the palette. Especially visible in the house image example. What is done here is done on micro levels of the signal.
Regarding the cast - this solution is greyscale neutral. What it also does is make tonalities visible which typically digital cameras "choke". Such as those close to white. There is no pure white in object reflections in Nature, there is always some very subtle cast close to it, depending on light properties and ambience reflection, which the observer's mind compensates. In the house example you may notice the warm Sun cast on wooden surface and on the bark of the tree, which is not visible in the first example but makes the image come alive. Taking care of those micro levels is very tricky and takes time and Helium treatment is still in the works.
Also, these solutions are designed to take care of the fixed offsets of the signal by measured reference behavior of the camera, not to compensate for unknown variables of light, that part should be done manually in post, and once the signal is treated that becomes much easier. Touch of white balance compensation and things like these tend to sink in.
- Thread starter
- Banned
- #55
Hrvoje Simic
Banned
Dragon.
Jacobo Martinez
Jacobo Martinez
- Banned
- #56
Mike Nagel
Banned
last one Epic Dragon or Weapon Dragon ?
- Thread starter
- Banned
- #57
Hrvoje Simic
Banned
Dragon.
Luigi Valtulini, Cine5K
Luigi Valtulini, Cine5K
- Thread starter
- Banned
- #58
Hrvoje Simic
Banned
Same solution on all Dragon imagery. Same WB & ISO.
No creative grading.
First images: RG3, DC2 out of the box
Second: Log3G10, RWG + custom treatment based on sensor & signal analysis
Differences range from subtle to drastic, manifesting throughout various scenes depending on signal.
No creative grading.
First images: RG3, DC2 out of the box
Second: Log3G10, RWG + custom treatment based on sensor & signal analysis
Differences range from subtle to drastic, manifesting throughout various scenes depending on signal.
Anton Shavlik
Well-known member
Those last few convinced me to buy whatever you're selling.
John Schick
Well-known member
Hey, these are some interesting developments so far. From looking at this example, the tattoo seems a little too green compared to the pink skin (in my opinion). Possibly the yellow/green adjustments could be dialed in some more:
As a comparison, here’s an alternative rendering using the RED to Alexa LUT:
As a comparison, here’s an alternative rendering using the RED to Alexa LUT: