Which functions in Raw Processing Apps are actually Post Processing?

[obsolescence]

I know the advantages of working from Raw files, but different Raw processors have many features with their strengths and weaknesses in performing specific functions (such as CA correction). I'd like to find info on which functions are done purely at the Raw processing stage (Noise Reduction?), and which ones are actually Post processing that might be better done in a different application (sharpening? perspective correction?).

 

[Iliah Borg]

I know the advantages of working from Raw files, but different Raw processors have many features with their strengths and weaknesses in performing specific functions (such as CA correction). I'd like to find info on which functions are done purely at the Raw processing stage (Noise Reduction?), and which ones are actually Post processing that might be better done in a different application (sharpening? perspective correction?).

Operations that depend on neighboring pixels can't be easily done at Bayer raw data level. The moment an operation involves neighboring pixels the maneuver is very limited because such operations affect raw data spatially, thus some resolution can be lost and some artifacts can be introduced above the level considered normal for the given operation. Such operations need to be very mild, or they are better done after demosaicking.

You can perform dot-noise reduction because it is pixel-level operation; it's much harder to apply any noise reduction that meaningfully reduces blotches. Instead, you can apply a very specific mild sharpening to Bayer raw data to prevent forming blotches during demosaicking.

To prevent amplification of artifacts such as CA during demosaicking, full or some CA reduction is often applied before it.

Another kind of operations that is impractical to apply to Bayer raw data: operations that split a pixel value creating additional colour components, like R or G values for a B pixel. An example would be applying a non-diagonal colour correction matrix, while white balance is can be applied as a diagonal matrix, and is often (but not always) performed at Bayer raw level.

Generally, it's a kind of art, one can mix and match, but mostly nobody bothers because higher megapixel counts are cheaper than this kind of research.

 

[obsolescence]

I just caught this. Thanks for your detailed response, Mr. Borg.

 

[brucet]

Technically anything that 'changes' the RAW file would be post processing. Yes no?

regards

 

[Iliah Borg]

Technically anything that 'changes' the RAW file would be post processing. Yes no?

regards

Raw to full-colour bitmap is conversion, anything else is just normal image editing. Image post-processing is image adjustments that enhances quality of specific output. For example, this is often done at prepress.

 

[Iliah Borg]

I just caught this. Thanks for your detailed response, Mr. Borg.

Happy if I managed to make it a bit clearer

 

[obsolescence]

Technically anything that 'changes' the RAW file would be post processing. Yes no?

regards

Raw to full-colour bitmap is conversion, anything else is just normal image editing. Image post-processing is image adjustments that enhances quality of specific output. For example, this is often done at prepress.

This may be subjective, but the terms "editing" and "conversion" and "processing" and "retouching" have fairly distinct meanings to me. Editing is the evaluation, selection, and organization of images. Raw conversion is somewhat limited to the interpretation of available data from the camera mapped to a color gamut, preferably without losing image detail or causing excessive noise. Post Processing is the manipulation of the available image information as an expressive interpretation of what was captured. Retouching is the repair of perceived defects in the subject or the capture system. Profile conversion is the remapping for a different output. Of course there will be overlaps, and software programs typically incorporate a range of these functions.

It would be nice if all my favorite implementations of the functions were available in one Raw Processing Application, but that hasn't happened yet AFAIK. Certain applications are superlative in a specific area, such as PTLens for distortion correction of non-native lenses (several of which are important to me). RawDigger is indispensible for evaluating exposure and learning the limitations of my digital capture system. Lightroom does CA correction best. Photoshop is unbeatable for all-around processing. DxO Photolab has superb noise reduction and perspective correction. Topaz Detail gives me the sharpening control I like.

Experimentation is the best way to find the best sequence of steps for a particular job, but now I'm confident that once I have done a proper Raw conversion to a 16-bit image, I probably won't be degrading it noticeably by doing the other steps in other software -- it will just take longer. The two exceptions to this are CA correction and NR, which (as I thought you acknowledged) should at least partially be done during Raw conversion. If I add more contrast/saturation in Post and then artifacts like banding in the sky show up, maybe I'll start over with a more extensive (and potentially destructive) Raw processing to get closer to the desired result.

 

[Doug Haag]

Technically anything that 'changes' the RAW file would be post processing. Yes no?

regards

Raw to full-colour bitmap is conversion, anything else is just normal image editing. Image post-processing is image adjustments that enhances quality of specific output. For example, this is often done at prepress.

This may be subjective, but the terms "editing" and "conversion" and "processing" and "retouching" have fairly distinct meanings to me. Editing is the evaluation, selection, and organization of images. Raw conversion is somewhat limited to the interpretation of available data from the camera mapped to a color gamut, preferably without losing image detail or causing excessive noise. Post Processing is the manipulation of the available image information as an expressive interpretation of what was captured. Retouching is the repair of perceived defects in the subject or the capture system. Profile conversion is the remapping for a different output. Of course there will be overlaps, and software programs typically incorporate a range of these functions.

It would be nice if all my favorite implementations of the functions were available in one Raw Processing Application, but that hasn't happened yet AFAIK. Certain applications are superlative in a specific area, such as PTLens for distortion correction of non-native lenses (several of which are important to me). RawDigger is indispensible for evaluating exposure and learning the limitations of my digital capture system. Lightroom does CA correction best. Photoshop is unbeatable for all-around processing. DxO Photolab has superb noise reduction and perspective correction. Topaz Detail gives me the sharpening control I like.

Experimentation is the best way to find the best sequence of steps for a particular job, but now I'm confident that once I have done a proper Raw conversion to a 16-bit image, I probably won't be degrading it noticeably by doing the other steps in other software -- it will just take longer. The two exceptions to this are CA correction and NR, which (as I thought you acknowledged) should at least partially be done during Raw conversion. If I add more contrast/saturation in Post and then artifacts like banding in the sky show up, maybe I'll start over with a more extensive (and potentially destructive) Raw processing to get closer to the desired result.

I'll be interested to hear the answer. But I presumed that what Iliah meant by "image editing" is what you have referred to as "Post Processing". And that would involve manipulation of the data resulting from the raw conversion process but not the activities involved in that conversion step itself.

 

[obsolescence]

Technically anything that 'changes' the RAW file would be post processing. Yes no?

regards

Raw to full-colour bitmap is conversion, anything else is just normal image editing. Image post-processing is image adjustments that enhances quality of specific output. For example, this is often done at prepress.

This may be subjective, but the terms "editing" and "conversion" and "processing" and "retouching" have fairly distinct meanings to me. Editing is the evaluation, selection, and organization of images. Raw conversion is somewhat limited to the interpretation of available data from the camera mapped to a color gamut, preferably without losing image detail or causing excessive noise. Post Processing is the manipulation of the available image information as an expressive interpretation of what was captured. Retouching is the repair of perceived defects in the subject or the capture system. Profile conversion is the remapping for a different output. Of course there will be overlaps, and software programs typically incorporate a range of these functions.

It would be nice if all my favorite implementations of the functions were available in one Raw Processing Application, but that hasn't happened yet AFAIK. Certain applications are superlative in a specific area, such as PTLens for distortion correction of non-native lenses (several of which are important to me). RawDigger is indispensible for evaluating exposure and learning the limitations of my digital capture system. Lightroom does CA correction best. Photoshop is unbeatable for all-around processing. DxO Photolab has superb noise reduction and perspective correction. Topaz Detail gives me the sharpening control I like.

Experimentation is the best way to find the best sequence of steps for a particular job, but now I'm confident that once I have done a proper Raw conversion to a 16-bit image, I probably won't be degrading it noticeably by doing the other steps in other software -- it will just take longer. The two exceptions to this are CA correction and NR, which (as I thought you acknowledged) should at least partially be done during Raw conversion. If I add more contrast/saturation in Post and then artifacts like banding in the sky show up, maybe I'll start over with a more extensive (and potentially destructive) Raw processing to get closer to the desired result.

I'll be interested to hear the answer. But I presumed that what Iliah meant by "image editing" is what you have referred to as "Post Processing". And that would involve manipulation of the data resulting from the raw conversion process but not the activities involved in that conversion step itself.

"Manipulation of the data" IS the conversion, so I think you meant "manipulation of the image information" after conversion (yes, that's what I call Post Processing). Judging from results in Lightroom, I think CA is most effectively done during Raw conversion, otherwise it can leave telltale gray edges; Iliah Borg's statement was significant: "To prevent amplification of artifacts such as CA during demosaicking, full or some CA reduction is often applied before it." I think some Color NR during conversion is useful as well (not Luminance). I think some NR that is programmed into the Raw processor is implemented after the conversion, but standalone applications offer greater control.

So, I start with RawDigger to evaluate the Raw files and select the ones to convert. I use the Raw conversion with minimal tonal & color adjustments but complete CA correction if possible (not possible with shifted image), Color NR if needed, distortion correction of native lens if available, and maybe perspective correction, CA correction of shifted image after distortion correction. Non-native lens (especially shift) gets corrected with possible enlarged canvas in PTLens (standalone or PS Plug-in) before minor perspective correction. Then Photoshop global & local adjustments and minimal sharpening, downsizing-sharpening and/or conversion to 8-bit sRGB for Web.