I am a renderer and use images to be seen on the screen, I do not need more than sRGB. I wonder if it is correct to use Prophoto to edit having more colors, that is, more information to avoid losses in the settings, while I see the scrren with proof colors in sRGB mode and at the end of the process convert the profile to sRGB to save the PNG? (all on 16bits)
In other words, for the internal calculation of Photoshop, it is the same to make an adjustment in Prophoto and then convert to sRGB for export, than make the adjustment directly in sRGB (maintaining 16 bits in both cases)? or making it with sRGB have color losses because of the calculation?
The Prophoto working colourspace has various disadvantages, you already know it's 16 bit only (because its so massive) - it also potentially contains significant data that cannot be displayed properly on a monitor display screen.
Next - the conversion from Prophoto to sRGB isn by necessity, a colorimetric rendering (no matter what Photoshop seems to offer), so it can have none of the promised (but not always functioning as expected) gamut mapping features of Perceptual rendering.
In a nutshell, this means that during the conversion from Prophoto to sRGB colours that fit within the destination colour space volume will not be desaturated, any colour values within Prophoto, but outside sRGB, will be mapped to "the skin" of sRGB, basically, this means that detail in high saturation areas can be lost - since previously varying values sitting along the same hue angle are mapped to a single value.
Its called 'clipping' and can be seen in the histogram post conversion.
I hope that aids your decision making?
I hope this helps
if so, please "like" my reply and if you're OK now, please mark it as "correct", so that others who have similar issues can see the solution
neil barstow, colourmanagement.net
[please do not use the reply button on a message in the thread, only use the one at the top of the page, to maintain chronological order]
ProPhoto is only useful in a print workflow where the output device gamut includes color that is outside of the sRGB space, which can be a considerable percentage of a typical CMYK space. Even then you probably wouldn’t want to use ProPhoto in a lowend print workflow that goes directly to press with no contract proofing or understanding of the destination CMYK output numbers. ProPhoto would have no advantage when the only destination is a web page.
In fact ProPhoto has very few advantages, except in a very intermediate editing stage where you need to contain certain extremely saturated colors to avoid premature clipping - until you decide what to do with them.
Lightroom/ACR is a case in point. Very often you get processing artifacts in the form of extremely saturated patches. That doesn't necessarily mean those colors are "real" or even worth keeping. But you don't want to just hard clip them, that doesn't look good. You need to remap.
Further down the line, as you get closer to a specific output, ProPhoto just causes a lot of problems. You still have all these saturated colors that you need to remap for output. The less remapping you have to do, the easier your life gets.
It is often argued that the bigger the gamut, the better. Hence ProPhoto is supposedly always "better". I strongly disagree. Good color has nothing to do with absolute saturation. It's about building relationships. Painters have made great works with pigments of limited purity throughout history.
ProPhoto is a useful tool in some very specific scenarios. That doesn't mean it's always the best tool for any job.
One very specific practical problem with ProPhoto is the extremely compressed shadow values (huge gamut means something else has to go. Gamma 1.8 exaggerates it). That means subtle shadow adjustments are very difficult.
I strongly disagree. Good color has nothing to do with absolute saturation. It's about building relationships. Painters have made great works with pigments of limited purity throughout history.
There are plenty of examples of the highly saturated great works—Sol Lewitt, Frank Stella:
There’s nothing about ProPhoto that stops you from working within a limited color palette—with sRGB that’s the only choice.
"There’s nothing about ProPhoto that stops you from working within a limited color palette—with sRGB that’s the only choice."
No, but then you're stuck with all the disadvantages of ProPhoto. And there are many.
There's an interesting concept called Pointer's gamut:
"The Pointer’s gamut is (an approximation of) the gamut of real surface colors as can be seen by the human eye, based on the research by Michael R. Pointer (1980). What this means is that every color that can be reflected by the surface of an object of any material is inside the Pointer’s gamut."
Pointer's gamut more or less coincides with Adobe RGB. If you looked for it I'm sure you could find a very few colors outside Adobe RGB, the question is how significant that is. Of course transmitted light can have higher saturation - but Frank Stella and Sol Lewitt (both of whom I hold in high regard BTW) should fit well within Adobe RGB. Colorimetrically speaking, of course 😉
If you have a wide gamut monitor and turn saturation all the way up - do you really feel you have "not enough color"?
Hello everyone and thanks for the replies. I do not need to know what Prophoto is used for, I just want to understand the internal operation of photoshop according to the color space, to be able to make informed decisions. I just asked if working with an image, for example applying saturation + curves, how is the internal calculation of photshop?
Option 1: Working with Prophoto I decide to apply infinity of curves, contrasts and saturations to edit an image. As prophoto has more colors, it will have more possibilities to distribute the colors resulting from the calculation of the curves. Then I will convert to sRGB in perceptual mode and send it to the client.
Option 2: Working in sRGB I will apply the infinity of cruvas and saturations doing exactly the same edition as before and send it to the client.
My question is whether with option 1 do I get exactly the same as with option 2? Is it true that with option two the margin of error is added after each calculation of curves and the final result of the relationship between the colors is poorer than having used prophoto?
ProPhoto wouldn’t expand the final sRGB space, so there would be no benefit in working in the larger space, and then converting. 16 bit sRGB would be the best you can do.
Hey people. My problem is not visualization. I could do the same theoretical exercise without a screen. I am asking about Photoshop's internal calculations, not its viusalization. The visualization during the work will be with proof colors in sRGB despite using Prophoto and the final viusalization will be in sRGB so it does not matter if my screen does not even reproduce adobeRGB. Nor does it make sense to think of painters, they did not use bits or make calculations between bits to obtain color. They had infinite bits to work and never had banding.
We imagine that for a pixel value the distance between two greens is 10 shades of green. If I saturate the image I increase the distance from 10 to 20 green tones. Among these 20 shades of green there are 20 different green colors making the progression. If I did the same in sRGB when it increased from 10 to 20, sRGB could only provide me with 5 new more greens (its an imaginari example, not the reality). So when I increase the saturation from 10 to 20, sRGB could only fill 5 green interleaved instead of 10. The other 5 spaces would be interspersed without color because sRGB has no more colors to interpolate. So when I move from Prophoto to sRGB I am interpolating his 10 new colors to 5 colors and i finish with only 5 colors like if I had done it from the beginning with sRGB. The question is: is it tighter to make perceptual compression from 10 colors in Prophoto to obtain 5 colors in sRGB, or is it better to generate 5 colors directly in sRGB?
I think it is more tight to compress 10 colors because you can choose which colors compress more, than not to generate 5 colors in a more limited color space that does not allow you to choose between 10 to decide which of the 5 suits you best.
What do you think?
A ProPhoto image could have the same appearance as an sRGB image. Here the image on the left was converted to ProPhoto from the sRGB image on the right. As expected they have different histograms despite the same appearance
If the ProPhoto image gets converted back to sRGB the histograms are identical, so there would be no color advantage in working in ProPhoto when the destination is sRGB. If you are worried about gray levels then it’s the bit depth that matters not the larger space:
Para hacer esta comparativa de una forma coherente con mi pregunta deberias mostrar una imagen retocada con varias capas de curvas en prophoto, y convertida en sRGB. Entonces compararla con otra con exactamente los mismos retoques hechos en sRGB. Habria que buscar una imagen que tuviera originalmente varios colores al limite del sRGB para que cuando se le hagan los retoques algunos colores crezcan a Prophoto mientras seguimos haciendole retoques, entonces si tendria sentido la comparación.
General rule of thumb: work in the smallest color space that will contain everything you need to do. That gives you the best precision. The larger the color space, the lower the precision, all else being equal. Each numerical step has to cover more ground.
If the target here is very specifically sRGB, it makes no sense to work in anything other than sRGB from the start.
It would be different if your target was a master file that could be used for different unspecified output. Then you want a color space big enough to cover most realistic scenarios - if the content requires it.
Again, as I said above, extremely large color spaces can be useful for editing headroom. They don't provide any advantages on their own - quite the contrary, there's usually a price to pay. And once you're in that large color space, it can be tricky to get out again.
You say that in another forum:
"The main advantage of ProPhoto is that it gives you headroom in the editing process. You can work without running into gamut clipping from which you can't recover. Once a channel is clipped, that information is lost."
So i take out my render from Corona renderer and 3dsmax (they work with widegamut) with 32bits EXR file and I introduce it in photoshop. At the time to introduce it I have to select an space color to work in photoshop and then i have to convert it to 16bits doing a tonemap highlight compresion. If i do that in sRGB I'll clip and lose a lot of colours as you said. So it is possible than can be better do all that in Prophoto and convert it to sRGB at the finish?
Yes, and that's the point I've been making all along: you want to avoid clipping. The longer you work in ProPhoto, the greater the risk of clipping in the final conversion. Put another way: ProPhoto just postpones the problem. You'll get there sooner or later.
But as an intermediate container, ProPhoto can be useful - for instance in your specific case here. But you'll need to remap the out-of-gamut areas, and the longer you wait, the harder it gets. IME it's best to tackle it as quickly as possible.
Clipping isn't always bad. It obviously depends on the type of image. I'm speaking of photographs mainly, where clipped areas tend to take on a dense, opaque quality lacking in air and natural light. It just looks heavy-handed. And sometimes that can even be a desired quality - there are photographers who use clipping to great effect.
Anyway, soft-proofing will tell. Provided you have a monitor with a gamut wide enough to cover the target color space. The histogram is also an underrated tool. If you see something's wrong, the histogram can usually tell you exactly where the problem is.
(Edited post morning-coffee 😉 )
And if we take it further, as happens in image editing, we not only saturate once but thousands of times saturate and desaturate, we use curves for everything. That means that these 5 colors that the sRGB had generated me when making changes again would happen to me from 5 to 3 because sRGB would not have where to take them, and when making more changes I would spend 3 to 1. doing the same thing in Prophoto would start with 10 (not with 5) and maybe it would end with 6. Then it is much more faithful to compress 6 colors to 5 (prophoto to sRGB) than to stay in sRGB with 1 color repeated 5 times. it's not like that?
ProPhoto maintains more color information during operations than sRGB. It is lije making calculations with sone decimals instead of integers.
If you work without decimals, you round numbers at evey step (like divisions) and inverse ooerations will have lost the decimals. Working with decimals bring them back and you round only the final result. In integers, 25/100 = 0, with decimals, .25. If you later remultiply by 10, the result will be rounded at 3, much better than 0.
same goes with bit depth and colour calculations.
No. It's the opposite. Within the gamut it covers, sRGB has higher precision than ProPhoto. Each numerical step, 0 - 255, has to cover much more ground in ProPhoto. The same number of steps in the ladder, but the distance between them is much greater in ProPhoto.
That's the price you pay for the large gamut.
Podrias poner un ejemplo como Jaques Archambault ha hecho con el ejemplo de los decimales? porque no te consigo entender. Cada paso numerico de 0-255 tiene que cubrir mucho mas terreno y por esto el calculo interno tiene mucho mas terreno para calcular, no es así? entonces puede calcular con mas terreno y sacar resultados mas exactos. Hay muchos mas decimales entre cada paso de la escalera, y el precio que pagas es que tenga que calcular mucho mas, pero el resultado es más preciso porque hay mas colores para elegir la precisión, no es así? Eso solo para los calculos, luego al finalizar todas las aplicaciones de curvas y demas, lo convertimos a sRGB para la entrega final.
Isn't that a description of 8 bit colour? If so, any limitation of this kind would be negated by the use of 16 bits (or more) per pixel.