Can someone give me the technical and/or practical reasons as to why many of the Artistic filters are not available and are "grayed out" if an image is in 16 bit mode?
Because they are 8bit!, but I think you knew that already. 😉 In simple words: Most of these filters are a combination of threshold, edge detection, dilation and blur operations, with some Fractal Noise thrown in here and there and there is no benefit in having them run in 16bit, as in most cases they will completely overwrite the previous pixel values. It would be possible, though.
Also, do you really gain anything by staying in 16 bit or lose any quality converting to 8 bit?
It all depends on what you do. Higher bit depths provide greater precision which is in particular relevant for photo touch-up and color corrections. Depending on the resolution of your image it also affects smoothness of gradation for gradients, blurs and so on, which is a simple math thing – when the distance spanned by a gradation is greater then the number of available pixel values, not every pixel can have a unique color any more. Therefore neighboring pixels may be assigned the same colors (banding) or interwoven with alternating pixel colors (dithering). Once that has happened, there is no way to restore the original smooth gradients, so you may wish to keep that in mind…
Mylenium
there and there is no benefit in having them run in 16bit, as in most cases they will completely overwrite the previous pixel values. It would be possible, though.
OK then, this partially explains it, but if it is possible to make these filters work in 16 bit, then why don’t they? I always assume that if something in PS is a certain way, then that is how Adobe intends me to do it. But the "8 bit only" filters thing seems to be at odds with keeping images at their highest quality throughout the work flow.
Didn’t you understand Mylenium’s explanation?
If the filter is going to mess around the pixels, what’s the point of previous operations being in 16 bit?
The use of 16 bit is of limited value to all but a few who understand why they are using it and for what advantage.
Furthermore, there aren’t many printers out there that can handle 16 bit.
Didn’t you understand Mylenium’s explanation?
No. That’s why I asked for further clarification. Didn’t you read my reply?
If the filter is going to mess around the pixels, what’s the point of previous operations being in 16 bit?
Well if you apply a filter as a Smart-Object, then it’s not really messing with original pixels, now is it?
The use of 16 bit is of limited value to all but a few who understand why they are using it and for what advantage.
Well those few who understand had to get their learning somewhere, now didn’t they? I had hoped this forum could be mine. Your snarky reply is neither helpful nor appreciated.
Use smart objects to run a filter on a 8 bits layer that resides in a 16 bits files.
Mostly because there’s not enough demand to spend the time converting the filter and testing it. If there is demand for a particular filter, then it’ll get converted. Making things function in 16 bit and 32 bit is work, then testing them is more work.
And yeah, there’s no real benefit to running them in higher precision.
Chris pretty much answered the question for you. However, I do see a need to have things work in 16-bit no matter what they do to an image. Adobe has been pushing non-destructive editing down everyones throat and have now gotten most of us hooked on it. Adobe has done something similar by pushing raw, we are now so used to being able to have so much more control in bringing out minute details in our images that we want that all of the time. Having to convert an image to 8-bit just to apply a filter or something flies in the face of this. If I want to work on my images in 16-bit then I want to be to do everything in 16-bit and not have to throw data away by converting to 8-bit. I also don’t want to waste time and space storing an almost complete 16-bit version and a completed 8-bit version.
This being said I am more than happy to cut Adobe some slack on the state of filters in Photoshop. There are far more important things that needs to be done to most of the filters in Photoshop than having them work in 16-bit. Most of the filters in Photoshop are now approaching the Jurasic area and go pretty much ignored when it comes to Photoshop updates. Things like being able to render a lens flare on an empty layer, more lens flare options, a much more sophisticated lighting filter, etc. are all more important things.
The simple solution to all of this is deal with it. Adobe does what Adobe does that’s it.
Robert
As you may guess, 16 bit data is twice the size of 8 bit data. This means the arrangement of this data in memory, as well as the instructions needed to manipulate the data, are different. The filter software must be designed to accommodate this difference.
Historically, many filters were created when Photoshop supported only 8 bit channel data. These filters were not updated when this situation changed.
Though the difference can be significant in certain situations, such as black and white or ProPhoto RGB images, for the vast majority of photographs, 8 bits works equally as well as 16 bits.
OK then, this partially explains it, but if it is possible to make these filters work in 16 bit, then why don’t they?
Well, you sure have a point, but as Chris said – even an ever so slight change in a feature needs coding work and testing and it would seem, that the demand for the artsy filters working native in 16bit is not that high to justify the effort. I would not take this as "Adobe wants me to work this way", though. As someone originally coming from 3D programs and compositing with After Effects, I can find simialr examples everywhere. The point really is, that it does not make sense to convert all types of effectsas to higher bit depths. Toon shaders in 3D programs are still mostly 8bit for apparent reasons and likewise, there would be no visual difference calculating a circle effect in After Effects in anything other than 8bit. If you want to attribute a failure to Photoshop, then it is its inability to mix different color depths easily. Still, not all is lost. If you encounter such situations, you can always duplicate the image, reduce the bit depth or change color space, run your filters and then copy&paste or drag the processed layer back to the origianl document. A similar workflow could be established by using smart objects, though then it will be a bit more tricky dealing with color profiles and such.
Mylenium
