Luminous Landscape Forum
Raw & Post Processing, Printing => Digital Image Processing => Topic started by: billbane on February 12, 2019, 05:04:09 pm
-
Anybody tried this?
So far, I have only seen something between "no change" and minor degradation via loss of resolution/muddying.
Is it possible that this is just for Fuji xtrans folks, at least as of now?
-
There are several articles posted explaining, this one does a good job:
https://laurashoe.com/2019/02/12/enhance-raw-file-details-in-lightroom-classic-cc-and-lightroom-cc/?fbclid=IwAR2w0YBsNi0R7ymp2NqjtyMfBCuTCKfWXyhcV-I963QL59V8K6EEbbmaY0I
My experience with Panasonic cameras on very few images has been no change.
Richard Southworth
-
Enhanced Detail will range for doing absolutely nothing visually to producing in best case (according to internal Adobe testing with Siemens Star resolution charts) 30% increase of resolution (NOT increase of pixels!). There is of course a before and after preview and the ideal zoom ratio to examine whether it's worthwhile converting the data to a linear DNG.YMMV.
-
Have now tried Nikon D850 and D90 DNG and NEF. In each case, take to Photoshop, as two layers, setting top one to "difference".
Photoshop shows no difference.
I gotta be missing something.
-
Should have mentioned that I extracted the respective two green channels and there was no diference between them either using Photoshop "difference" blend mode.
Something wrong with me or Adobe.
-
Have now tried Nikon D850 and D90 DNG and NEF. In each case, take to Photoshop, as two layers, setting top one to "difference".
Photoshop shows no difference.
I gotta be missing something.
A better way:
http://digitaldog.net/files/Apply_Image.pdf
-
Have now tried Nikon D850 and D90 DNG and NEF. In each case, take to Photoshop, as two layers, setting top one to "difference".
Photoshop shows no difference.
I gotta be missing something.
"AI" is saying you don't need sharpening... :)
-
Tested on a hand-held OLY OMD EM1-Mk2 shot of a mediaeval retable (altar piece). Shot with an OLY 12-100 f4 lens @18mm, ISO6400. Lots of intricate detail in the carvings, a little bit noisy but acceptable and sharpish. 21mb after processing in LR.
After 2 minutes of additional processing, the "Enhanced" DNG is 108mb (!) and not an iota of visible difference even when pixel-peeping.
I wish I'd known at the time I shot this what you already know: that if I'd taken say 16 images on a high-sequential shutter setting and merged them in P'shop using 'mean' layer blending, I would have had a much less noisy image (& with much more detail if I'd avoided movement).
-
Good to have better edges, less moiré and more small details…
Some examples where it made a difference here (https://i.imgur.com/U6mA8c0.gif), here (https://i.imgur.com/b0pqQSW.gif) and here (https://i.imgur.com/hdHHGGX.gif). (Animated GIF)
RAW files used:
https://www.imaging-resource.com/PRODS/D700/D700hSLI00100.NEF.HTM
http://download.dpreview.com/fujifilm_xpro1/DSCF2008new.RAF.zip
https://www.imaging-resource.com/PRODS/sony-a7r/AA7RhSLI00050NR0.ARW.HTM
-
After 2 minutes of additional processing, the "Enhanced" DNG is 108mb (!) and not an iota of visible difference even when pixel-peeping.
That's WHY there's a very quick (at least for me) preview so you can decide or not to move forward on an image by image basis. And yeah, it's a lot larger; it's linear RGB data now. Nothing new in that respect.
-
Pictus,
Thanks for the 3 different test cases with 3 different cameras. I thought the changes were very subtle, except the N700 where the color repair was noticeable.
I am hoping they have bigger plans for this since as of now it looks very niche oriented. Their placement in their menu structure suggests something with substantial and general applicability.
Based on your trials, can you formulate the balance of the statement, "I would consider using Adobe's enhance details...."?
-
Pictus,
Thanks for the 3 different test cases with 3 different cameras. I thought the changes were very subtle, except the N700 where the color repair was noticeable.
I am hoping they have bigger plans for this since as of now it looks very niche oriented. Their placement in their menu structure suggests something with substantial and general applicability.
Adobe DOES have bigger plans for this.
-
I have found that the new Lightroom Classic Enhanced Details function, via my Win 10 Pro v1809 64bit, Intel Core i7-4790 @3.60Ghz, 12 GB RAM PC, took 2 minutes 45 seconds for LR to create the Enhanced version of a raw photo taken with a Fuji GFX 50R and GF 45mm lens at f5.6. The photo has a significant amount of moire in distant fences and foliage. The Enhanced version has significantly reduced moire with much of it eliminated entirely.
For images with significant moire the Enhanced Details function will be of significant benefit to me because the alternative, using the moire brush to eliminate pervasive moire in an image, is very tedious and will often result in making that raw image very slow to load and to work on.
Here are some screen shot comparisons with the raw on the left and the enhanced version (DNG) on the right of each comparison screen shot. The view is 4:1 in LR to make seeing the differences easier.
Dan
-
Pictus,
Thanks for the 3 different test cases with 3 different cameras. I thought the changes were very subtle, except the N700 where the color repair was noticeable.
I am hoping they have bigger plans for this since as of now it looks very niche oriented. Their placement in their menu structure suggests something with substantial and general applicability.
Based on your trials, can you formulate the balance of the statement, "I would consider using Adobe's enhance details...."?
You are welcome.
As already mentioned, it is excellent for moiré, check
https://www.dpreview.com/forums/post/62299391
As a pixel peeper I always will use... :)
-
... Win 10 Pro v1809 64bit, Intel Core i7-4790 @3.60Ghz, 12 GB RAM PC, took 2 minutes 45 seconds for LR to create the Enhanced version of a raw photo taken with a Fuji GFX 50R and GF 45mm lens at f5.6.
Note that the most important component for the "Enhanced Detail" function is the graphics card / GPU. Total RAM and processor are less important
-
I would like to see a GPU benchmark for "Enhanced Detail".
The new AMD Radeon VII is very good for compute
https://www.anandtech.com/show/13923/the-amd-radeon-vii-review/15
-
And why the file size increase?
I do understand that the linear gamma RGB data addition to the DNG container requires more space, but why the increase in the first place?
Why not an improved (if any) version of a better Raw conversion?
Cheers,
Bart
-
And why the file size increase?
I do understand that the linear gamma RGB data addition to the DNG container requires more space, but why the increase in the first place?
Why not an improved (if any) version of a better Raw conversion?
Cheers,
Bart
It produces a demosaiced RGGB raw file (yes, 4 channels), which in the typical case increases the size by 4x (now for every pixel you have RGGB values)
-
It produces a demosaiced RGGB raw file (yes, 4 channels), which in the typical case increases the size by 4x (now for every pixel you have RGGB values)
Hi Frank,
I understand that part, but why not improve the Raw conversion itself, instead of storing a post-processed Rawish version of it (as if it where Raw, which it apparently isn't)? Wouldn't an output option make more sense?
Cheers,
Bart
-
why not improve the Raw conversion itself, instead of storing a post-processed Rawish version of it (as if it where Raw, which it apparently isn't)? Wouldn't an output option make more sense?
Cheers,
Bart
Bart,
I don’t know enough to tell if what enhance details is doing can be achieved with an improved raw conversion, maybe it can.
However, besides the discussion if the enhanced DNG is raw or not, it is in a early stage of the processing pipeline. It is not color space encoded, which means you can apply a dcp profile and you can apply (or copy) any other adjustment as if it were an original raw file. None of this can be done with a rendered (output) file.
-
Why not an improved (if any) version of a better Raw conversion?
Cheers,
Bart
This.
If anything, AI should be extremely capable at demosaicing. In fact, it could be trained easily with the ibis four-shot hr images like Olympus e.a. offer.
Probably a well designed AI structure will also include additional processing that might improve pattern detail or noise reduction etc., so it's useful to offer this as separate options, but in the current offering it seems the results don't quite justify the effort.
-
If anything, AI should be extremely capable at demosaicing. In fact, it could be trained easily with the ibis four-shot hr images like Olympus e.a. offer.
Phase One did this already, years ago, using data from the Phase One P45+ and Phase One Achromatic (same sensor, optical path etc, just with and without RGB filter on the pixel wells). It's one of the ways they made large improvements to the per-pixel quality of the Capture One raw processing algorithms; no additional step of processing required.
From some initial testing of LR's Enhance Details tool...
Detail Enhancement: Meh. Not much improvement, and where there is improvement it's only because LR's starting point is pretty mediocre. In this regard Adobe has added a convoluted step to their workflow that gets their image quality closer (but still, to my eye, not as good as) C1's native raw processing algorithms. This is true of absolute detail and fine lines, but it's even more true of the overall feel of the detail, which feels more organic and natural to me in C1 (again, without the need for a separate processing step).
Color Moire: Meh. The effect is that of locally blurring the color. If an area moire'd with a bias toward red then Enhance Details reduces the speckling, but not the underlying bias. So things are still randomly red or blue or green tinted that should be neutral. This can save time vs manually retouching for a quick-dirty use but for any sort of final use you'd still often need to manually retouch the color.
Luminance Moire: Impressive! In some cases subject matter that was rife with moire is cleaned up entirely. In other cases the moire is still there, but with a slightly tidier look. In some cases the moire is not improved at all. Overall that's very impressive.
Our experience is that, as you pass ~30-50mp into higher resolution sensors (e.g. 80mp, 100mp, 150mp), the chance of moire falls through the floor (still technically can happen, but far more rarely). But since most of the world is still shooting cameras in the 10-50mp range and likely will be for many years to come, this improvement in ACR is very welcome and will be broadly useful.
Hopefully as computing power improves and the programming/frameworks behind this tool improve the effect can just be folded into the normal processing chain.
Bias Alert: I'm an admitted C1 fanboy and must be considered biased as my company sells P1 gear.
-
Hi Doug,
Your post is confusing to me. Does C1 offer the features you list as options for enhancing details? As up to vr 11, I don't know of a enhance details settings. So is it something built in to the software? I know about the diffraction correction under the sharpness tab.
Just curious, as I am always looking for more details.
Thanks
Paul
-
Hi Frank,
I understand that part, but why not improve the Raw conversion itself, instead of storing a post-processed Rawish version of it (as if it where Raw, which it apparently isn't)? Wouldn't an output option make more sense?
Cheers,
Bart
It would make more sense if doable but today its not. Like its not doable to convert JPEG to raw.
-
Hi Doug,
Your post is confusing to me. Does C1 offer the features you list as options for enhancing details?...
Doug was talking about LR.
-
One area that I thought would immediately benefit from the claims being made about the AI detail enhancement would be the treatment of Canon DualISO images made with the Magic Lantern firmware hack. The images were produced by merging the simultaneous recording of a single scene capture read off the sensor in pairs of lines alternating between two user-defined ISO values. The resulting image essentially merges two captures spaced (ISO1-ISO2) stops apart, increasing the dynamic range of the capture, at the expense of linear pixel resolution. The result is a file that often displays CA and moiré, as well as aliasing and color artifacts along color edge contrasts - the exact list of things that this new feature is supposed to address.
I do not shoot Canon anymore, but I have an archive of 5DIII DualISO images - I ran 4 or 5 DNGs through the Detail Enhance procedure and the result is essentially the same as the input. I would have thought that the obvious artifacts would at least change a little and would be easy pickings for whatever AI algorithms are doing their thing during processing.
In terms of the x-trans images I have tested with it (from my X-H1) there is an observable difference, but it is really at the edges of details and you need to be zoomed into the image 200-400% to see any noticeable change. This may be of benefit if one intends to enlarge the image significantly. While the difference is observable, it does not appear to be significant.
Unless you make a business shooting Seimens stars, it appears that this feature is not particularly useful at the moment.
kirk
-
AFAIK, there are only two people here posting who had any experience, under NDA, speaking with Adobe engineers like Simon Chen and Eric Chan about the development of Enhanced Detail last year as it was being developed and tested. Here are some facts, as much as I can provide under NDA and the facts can be corroborated by other betas.
Adobe didn't wake up on day and thought, "let's produce a new rendering process that forces people to take the time to convert raw to linear DNG and extra space" because they felt those two attributes were good for customers workflows per se. They did it that way because that's the way, for now, the feature had to be implemented. The "why did Adobe do this" question was asked last year, by me for one! What happens in the future may change. Asking today why this was implemented as it was, especially by people who don't have a clue about the ACR processing code, marketing people who work for competing companies of Adobe who may or may not know how their own software code actually works, is as silly as me asking why my 2017 Mazda CX5 doesn't run on electricity or why my iPhone X doesn't receive G5 cellular.
Adobe didn't claim this feature is for everyone or works well on every image. They did testing with Enhanced Detail with Siemens Star resolution charts showing a 30% increase of resolution (NOT increase of pixels!). They didn't claim more, they didn't state this was true for all images and cameras that capture this data.
There is of course a before and after preview and the ideal zoom ratio to examine whether it's worthwhile converting the data to a linear DNG so users can see, on a case by case basis, if they wish to convert and use Enhanced Detail.
The same people asking why this is, are some of the same people who spent post after post, page after page defending Topaz Lab's claim they convert JPEG to raw and can edit a JPEG as if it were a raw without any evidence to defend those claims. Adobe can defend the claim that in some captures, Enhanced Detail will enhance the detail. As can users. They cannot "defend" why one must convert the data to a linear DNG to the degree they explained their processing to those under NDA. Nor can they defend that if you use LR or Photoshop to convert a wide gamut image to sRGB, you'll clip colors. That's how it works kids. Unlike Topaz, they will not use marketing shills to state they can produce stuff that doesn't exist and processing that can't be backed up.
Now this is a new feature and it will evolve. One issue is the amount of processing and OS support needed today (simply examine the OS requirements for Enhanced Detail to work, one OS being just released a few months ago). Few here have any experience producing software. I have a little. Yeah, it is possible that ED (Enhanced Detail for short) could be produced directly from the raw without a DNG intermediate but what if 8% of the user based had hardware support for it? Wouldn't fly well now would it. What happens in the future happens. TODAY, if you want to use ED, you convert to a linear DNG and if you're smart, you view the preview first instead of doing a batch convert. How you'll handle this in a year is your guess. Some of us will know before that.
-
It would be helpful if Adobe developed a metric, or set of metrics, that calculated some increase in detail or acutance, or decrease in moiré or aliasing, and provided that metric in the preview window. A false color preview of the areas most affected by the algorithm would also be nice. This way, the user could get a visual representation of a threshold level of improvement and evaluate the need to apply the algorithm in a more informed manner.
Kirk
-
FWIW, I'm seeing effects on my Fuji Xtrans files ranging from "not much difference" to "wow, that's amazing." Better fine detail, better color detail, no "worms" in foliage and natural areas.
It didn't occur to me until photodan suggested it in another thread, but I tried it on a file from my GFX50S and it handles moire quite well. Posted a sample in that thread (link (https://forum.luminous-landscape.com/index.php?topic=129073.msg1095714).)
Haven't tried it on Canon or Sony files yet.
Even given all the downsides (separate DNG, long time to create), I'm still pretty happy with this new feature.
-
It would be helpful if Adobe developed a metric, or set of metrics, that calculated some increase in detail or acutance, or decrease in moiré or aliasing, and provided that metric in the preview window. A false color preview of the areas most affected by the algorithm would also be nice. This way, the user could get a visual representation of a threshold level of improvement and evaluate the need to apply the algorithm in a more informed manner.
Kirk
It's image and capture (device) specific. So they can't, YMMV. What they did test and can report is what they produced on a capture with one system of Siemens Star resolution chart. No more, certainly less depending on lots of possible factories.
-
Interesting - so Adobe cannot compare the differences between the original raw and the processed Enhanced Detail linear DNG and report those differences in a meaningful way? Weird.
kirk
-
With the X-Trans/ Raw files from my X100F I see a 2nd effect (aside from a somewhat cleaner rendition at pixel level): The sharpening controls in the Detail tab work more in a traditional way now with the Enhanced.DNG, i.e. allowing a higher Amount setting compared to the original RAF.
So altogether it can make a quite significant difference.
Note that this can not be previewed
(with the Preview function of Enhance Details, before the Enhanced.DNG is created).
--
-
Interesting - so Adobe cannot compare the differences between the original raw and the processed Enhanced Detail linear DNG and report those differences in a meaningful way? Weird.
kirk
You can easily use Compare mode in LR and examine the Enhanced Detail version to the original. What else would one need?
-
I was thinking more along the lines of something meaningful when evaluating the preview - like a false color map of the "improvement" or at least the difference between the original and the enhanced version. If there were some quantitative way of evaluating the effect, that value, or range of values, could be correlated with some real observable difference in the final output (print, etc.) and then it could be used as a go/no-go threshold for evaluating whether to spend the extra time and expense of enhancing the base conversion. It would be helpful if I knew for my images on my printer, if the effect did not generate a score of 10 or greater, then it is not worth running the enhance algorithm for my print output. Or, maybe an "effective MTF" could be computed to demonstrate the gain from the enhance routine. Etc.
This would be different than, for example, knowing that my specific image contained moiré that I would need to treat with the enhancement, regardless of the overall metrics.
What would be nice as well would be a side app that would run through a batch of raw files and compute the various metrics along with a small false color preview of the spatial distribution of the metrics and output the results to an HTML document that would be sortable based on highest to lowest effect, on a per-metric basis. This way you could run that in the background on a batch of images, making evaluation of the effect more quantitative and efficient. You could even make the app sort the images, based on their scores, into folders, or marked with ratings stars, flags, etc. that would get imported into Lightroom.
Just thinking out loud.
kirk
-
I don't mind making qualitative rather than quantitative judgments about image quality. I do it every time a make an image and every time I process an image.
-
I suppose eyeballing it can work. But that begs questions like "why do we calibrate and profile our displays and scanners and printer-paper combinations?" Why do we proof under controlled lighting? If eyeballing it works, it works! But maybe it doesn't.
Here is a mock up of what I am describing. I took a shot of a tight patterned weave of nylon that I know will produce moiré when photographed at a right distance with a particular lens+sensor - in this case a Fujifilm X-H1 and the 16mm f/1.4. The attached images compare the original conversion in ACR and the sync'ed enhanced version. The moiré is visually reduced, for sure. But there are other areas of the image that do not appear to have received any enhancement that actually have received changes similar to the moiré areas.
Attached are the two images in split window view at fit to view and then at 100% zoom - the moiré artifact is apparent and tints the neutral gray material with a purplish hue.
-
Here are the L, a and b channel comparisons for the enhanced versus original images.
-
In each of the three channel comparisons, I copied the respective channels into an RGB image, performed an "Apply Image..." and chose the base image in Subtract mode, Scale 2, Offset 128. Then I applied a steep curve to the result to amplify the differences from neutral gray (no difference) and then applied a false color gradient map to give a sense of positive or negative deviation.
What these images tell me is that both the L and the color channels are receiving modification, but the b channel is receiving stronger modification than the a channel. In this case, that is already obvious because of the particular issue that I can see even without rendering a preview of anything, but in cases where it is not so obvious, these kind of aids might be helpful in evaluating an image prior to enhancement.
Also, in the below image, where I zoomed out to include the whole image, there are other areas of enhancement (shown here is the L channel) that are being shown by the false color rendering that do not become apparent until you really zoom in and toggle back and forth. Are these worthwhile? I don't know, but at least I have some data to evaluate and compare with whatever output I generate and can evaluate the effect uniformly over large batches of images or image types.
kirk
-
I suppose eyeballing it can work. But that begs questions like "why do we calibrate and profile our displays and scanners and printer-paper combinations?"
Why? Calibration is a process to take a device and place it into a desire condition. Like making a display better match a print. Displays are not stable devices so we recalibrate to put that desired calibration back into effect. We proof under controlled conditions for the same reasons. We calibrate an E6 film line to a desired condition and replenish (recalibrate) to maintain that condition. We profile devices to fingerprint device behavior. So calibration and profiling have nothing to do with this discussion.
IF you charge by the hour, or if you use OPM (Other People's Media), you don't need and may not desire a lick of color management (calibration or profiling).
-
In each of the three channel comparisons, I copied the respective channels into an RGB image, performed an "Apply Image..." and chose the base image in Subtract mode, Scale 2, Offset 128. Then I applied a steep curve to the result to amplify the differences from neutral gray (no difference) and then applied a false color gradient map to give a sense of positive or negative deviation.
What this tells you is there IS a difference and where in the image, and as the article provided states, when using the process described: Pixels that aren’t level 128 gray are different by the amount they depart from 128 gray.
-
What this tells you is there IS a difference and where in the image, and as the article provided states, when using the process described: Pixels that aren’t level 128 gray are different by the amount they depart from 128 gray.
Which is why it would be a really nice feature to have in the preview. I'm glad we agree.
kirk
-
Which is why it would be a really nice feature to have in the preview. I'm glad we agree.
kirk
What preview? The one in LR? Seems overkill. If you'd see a useful visual improvement in the ED preview, don't convert.
-
Phase One did this already, years ago, using data from the Phase One P45+ and Phase One Achromatic (same sensor, optical path etc, just with and without RGB filter on the pixel wells). It's one of the ways they made large improvements to the per-pixel quality of the Capture One raw processing algorithms; no additional step of processing required.
From some initial testing of LR's Enhance Details tool...
Detail Enhancement: Meh. Not much improvement, and where there is improvement it's only because LR's starting point is pretty mediocre. In this regard Adobe has added a convoluted step to their workflow that gets their image quality closer (but still, to my eye, not as good as) C1's native raw processing algorithms. This is true of absolute detail and fine lines, but it's even more true of the overall feel of the detail, which feels more organic and natural to me in C1 (again, without the need for a separate processing step).
Color Moire: Meh. The effect is that of locally blurring the color. If an area moire'd with a bias toward red then Enhance Details reduces the speckling, but not the underlying bias. So things are still randomly red or blue or green tinted that should be neutral. This can save time vs manually retouching for a quick-dirty use but for any sort of final use you'd still often need to manually retouch the color.
Luminance Moire: Impressive! In some cases subject matter that was rife with moire is cleaned up entirely. In other cases the moire is still there, but with a slightly tidier look. In some cases the moire is not improved at all. Overall that's very impressive.
Our experience is that, as you pass ~30-50mp into higher resolution sensors (e.g. 80mp, 100mp, 150mp), the chance of moire falls through the floor (still technically can happen, but far more rarely). But since most of the world is still shooting cameras in the 10-50mp range and likely will be for many years to come, this improvement in ACR is very welcome and will be broadly useful.
Hopefully as computing power improves and the programming/frameworks behind this tool improve the effect can just be folded into the normal processing chain.
Bias Alert: I'm an admitted C1 fanboy and must be considered biased as my company sells P1 gear.
I have been experimenting with the tool via Capture One.
1. Open With - Photoshop
2. Photoshop opens with the picture open in Camera Raw
3. Apply Enhance detail to the picture (before/after shows an improvement)
4. Click done
5. If you are in session mode the enhance DNG shows up in Capture One straight away, if in catalogue mode you have to import it.
I then compare the enhanced DNG and original RAW in Capture One and detail wise they are identical.
I am puzzled then I read that one of the reasons the enhance DNG is so big is that the original RAW is embedded in the Enhanced DNG. I have a feeling that Capture One is reading the original RAW (.RAF and .NEF in my case) file and not the linear etc etc DNG created by the tool.
Does anyone have any thoughts on this?
-
I am puzzled then I read that one of the reasons the enhance DNG is so big is that the original RAW is embedded in the Enhanced DNG.
AFAIK, not the original proprietary raw. In LR, you end up with the original and a new DNG named Enhanced. That new DNG has a Linear DNG without the enhancement and the linear DNG with the enhancement. LR/ACR understand the enhanced version and other products, like C1 can (if it's properly supporting DNG (?)) showing you the linear DNG without enhancement. That's why the size is so much larger.
-
I have a feeling that Capture One is reading the original RAW (.RAF and .NEF in my case) file and not the linear etc etc DNG created by the tool.
Does anyone have any thoughts on this?
Hi Dave,
AFAIK that is correct. The DNG is no longer a pure Raw file as defined in the Published Standard, it is in a proprietary Adobe format that for the moment is probably only decodable with recent Adobe applications. It, therefore, defaults to the unenhanced version. That might also mean that it is unreadable for older versions of Lightroom and ACR.
Cheers,
Bart
-
Thanks Andrew & Bart. I shall stick with pure Capture One 8)
Dave
-
I have a feeling that Capture One is reading the original RAW (.RAF and .NEF in my case) file and not the linear etc etc DNG created by the tool.
After further testing it seems that not only C1 but other tools including raw digger are reading the original Raw and not the "enhanced" (linear DNG) version.
I previously stated that the "Enhanced" DNG was RGGB (4 channels), now I realise that I cannot tell if that is the case, since that was the original Raw.
-
After further testing it seems that not only C1 but other tools including raw digger are reading the original Raw and not the "enhanced" (linear DNG) version.
I previously stated that the "Enhanced" DNG was RGGB (4 channels), now I realise that I cannot tell if that is the case, since that was the original Raw.
It's NOT the original raw. It's a DNG. The sensor raw data is "original" yes but the Metadata is converted which is what occurred whenever one converts an original proprietary raw to DNG. But this is NOT the same as embedding the actual original proprietary raw in a DNG as you can when doing a conversation to DNG in LR or using the Adobe DNG converter. Think of EN as having 2 DNGs. One only the latest LR/ACR can understand.
According to Simon Chan:
"Capture One or any other 3rd party DNG readers can only understand and read non enhanced details portion of the DNG (original raw bits), which remains intact through the enhanced details process. Only Lr/ACR currently understands the enhanced detail part of the DNG."
-
It's NOT the original raw. It's a DNG. The sensor raw data is "original" yes but the Metadata is converted which is what occurred whenever one converts an original proprietary raw to DNG. But this is NOT the same as embedding the actual original proprietary raw in a DNG as you can when doing a conversation to DNG in LR or using the Adobe DNG converter. Think of EN as having 2 DNGs. One only the latest LR/ACR can understand.
According to Simon Chan:
"Capture One or any other 3rd party DNG readers can only understand and read non enhanced details portion of the DNG (original raw bits), which remains intact through the enhanced details process. Only Lr/ACR currently understands the enhanced detail part of the DNG."
Thanks for the clarification
-
I played a little with the tool and as previously said the effect varies from nothing to something.
Most differences are on raf files but even there it's inconsistent, some photos where I would have expected a significant difference (lots of green) are not that affected.
On the ones where a difference is clear I also notice an increase in saturation in addition to the small increase in sharpness, possibly explaining better the overall improvement (greater difference between adjacent colors?). I think part the reason the effect if less than expected is because Adobe stepped up their xtrans files processing, I see much less artifacts these days.
-
There may be a flaw in my reasoning, but I do the enhancement, work on the enhanced file in the Lightroom develop module, synchronize the corrections with the original RAW file, in my case an ARW. I then take the dng into Photoshop as a smart object and finish it there. When I'm all done in Photoshop, I delete the dng enhanced file and keep the ARW. I can always re-enhance it from the ARW and I do not need to store the large dng. This way I'm not storing an extra unnecessary file. Does anyone see an error in my logic? If so, I'd love to know. I know disc space is relatively cheap these days but it still seems wasteful to store a file I'll not likely use again and can be retrieved if I do.
-
I thought of DNG to be an open standard, so every company should have acces to the enhanced DNG.
If not Adobe does something like every camera maker is doing. Or am i wrong?
Maybe it is just a matter of time for companies like Phase-one to read the enhanced DNG.
-
I thought of DNG to be an open standard, so every company should have acces to the enhanced DNG.
If not Adobe does something like every camera maker is doing. Or am i wrong?
Maybe it is just a matter of time for companies like Phase-one to read the enhanced DNG.
The results of the new processing (which is proprietary) to an enhanced DNG is a linear demosaic'ed DNG.
DNG is openly documented by Adobe. Not everyone follows the documentation correctly.
-
The results of the new processing (which is proprietary) to an enhanced DNG is a linear demosaic'ed DNG.
DNG is openly documented by Adobe. Not everyone follows the documentation correctly.
Do you say, third parties, like Phase, could read this enhanced DNG if they just want to?
and if it is demosaic'ed; how much does it differ from a tif 16 bit? Or is that a silly question?
-
Do you say, third parties, like Phase, could read this enhanced DNG if they just want to?
and if it is demosaic'ed; how much does it differ from a tif 16 bit? Or is that a silly question?
Still '16-bit' (high bit data) but not totally raw; partially rendered. See: http://www.barrypearson.co.uk/articles/dng/linear.htm
No idea what Phase can and can't do; in the past, their adherence to DNG was iffy at best.
-
Has anyone tried processing Enhanced Details on a 2018 MacMini? If so, what sort of times were you getting?
-
Still '16-bit' (high bit data) but not totally raw; partially rendered.
Not Raw at all, but demosaiced RGB. Or, as Barry Pearson states: "This RGB image data may have come from demosaiced raw image data, or from another source such as TIFF or JPEG or something else."
Similar to the DNGs that a program like Topaz JPEG to RAW produces, cooked linear RGB, wrapped in a DNG container.
No idea what Phase can and can't do; in the past, their adherence to DNG was iffy at best.
DNG is just a container. It depends on the data in that container and whether that is well documented. AFAIK, "AI Enhanced" is not publicly documented, so it has nothing to do with what Phase One can do.
Cheers,
Bart
-
Has anyone tried processing Enhanced Details on a 2018 MacMini? If so, what sort of times were you getting?
No but it works through the GPU- CPU is not used.
On my 250$ card i do about one d850raw in 5 sec.
-
No but it works through the GPU- CPU is not used.
On my 250$ card i do about one d850raw in 5 sec.
I am interested in replacing my 2012 Mac Mini. It says Enhanced Details it will take 6 minutes, but the end it doesn't process the whole image leaving large black squares. I don't mind waiting a little, but I'd like to have some idea that it will actually work on the 2018 Mac Mini.
-
Not Raw at all, but demosaiced RGB. Or, as Barry Pearson states: "This RGB image data may have come from demosaiced raw image data, or from another source such as TIFF or JPEG or something else."
Similar to the DNGs that a program like Topaz JPEG to RAW produces, cooked linear RGB, wrapped in a DNG container.
DNG is just a container. It depends on the data in that container and whether that is well documented. AFAIK, "AI Enhanced" is not publicly documented, so it has nothing to do with what Phase One can do.
Cheers,
Bart
I never said Phase was iffy with Enhanced DNG or could deal with it. Been iffy in the past long before Enhanced functionality existed in past experience. No idea if they have improved support from last time I used C1 and I asked Doug here, no answer from him or others.
-
I am interested in replacing my 2012 Mac Mini. It says Enhanced Details it will take 6 minutes, but the end it doesn't process the whole image leaving large black squares. I don't mind waiting a little, but I'd like to have some idea that it will actually work on the 2018 Mac Mini.
AHA - that is probably why they did not make it the new raw developing standard...
Problem with the mac mini is the basic GPU...
Must say what i see is an improving in small fonts at pixels size and less moiré- so very small details... In most images you would not see any difference, only if you search for it and be able to compare.
-
I am interested in replacing my 2012 Mac Mini. It says Enhanced Details it will take 6 minutes, but the end it doesn't process the whole image leaving large black squares. I don't mind waiting a little, but I'd like to have some idea that it will actually work on the 2018 Mac Mini.
This is very hardware-dependent, and I don't think you can get any assurance before testing it yourself, and with your raw files. I see the problem all the time with Fuji Xt2 files on my MacBookAir, which processes bigger Nikon D800 files just fine. My Windows PC processes both types of file. In any case, it's very hard to distinguish the benefit, so I wouldn't factor ED into your purchasing decision.
-
In any case, it's very hard to distinguish the benefit, so I wouldn't factor ED into your purchasing decision.
If that is the case, no real need to upgrade until LR no longer supports High Sierra.