Luminous Landscape Forum
Raw & Post Processing, Printing => Adobe Camera Raw Q&A => Topic started by: Chris Kern on March 10, 2021, 05:02:44 pm
-
I've done a little playing around today with the new "Super Resolution" feature (https://blog.adobe.com/en/publish/2021/03/10/from-the-acr-team-super-resolution.html#gs.v8kobh) of Adobe Camera Raw and I must say I'm favorably impressed.
I've been using Topaz Gigapixel for quite a while when I needed to enlarge images. The new ACR capability similarly employs machine learning to in effect recreate a larger image with comparable detail to the original—although, unlike the Topaz product, it is (currently, at least) limited to a 2X increase in linear dimensions.
I haven't had time to perform a very rigorous comparison, but Adobe's convolutional neural network model seemed to do as well as Topaz's on the first couple of images I tried. The Adobe conversion was also much faster on my Apple laptop (2020 MacBook Pro). I suspect some other forum participants have been testing this feature for Adobe and perhaps they can share their experiences.
As Eric Chan of Adobe notes in the essay at the link above, the new feature is "coming soon" to Lightroom.
Attached: (1) a JPEG made from a tight crop of a Nikon NEF file (D800E), (2) the same crop from a 2X enlargement produced from the raw file by ACR, and (3) the same crop from a 2X enlargement produced from a full-resolution TIFF by Gigapixel (which I used to "bake in" a couple of minor tone adjustments I had made to the image).
-
Thanks for posting.
-
And if so desired, you can load a TIFF or JPEG (etc) into ACR (not ACR as filter), at least I can on Mac, using the Format Menu set to Camera Raw. Super Resolution can be used then.
-
Here's another example, this time with a Fuji X-Trans (X-T3) file. Same three rendering methods.
With this image, the Topaz version appears sharper, but perhaps also noisier. (I haven't applied manual sharpening to any of the files; what you see is what I got from the original Lightroom export, and the respective ACR and Gigapixel enlargements.)
From a practical perspective—i.e., assuming the purpose was to prepare the image for printing—I doubt the difference between the 2X enlargements made with Adobe and Topaz products would be noticeable.
And, of course, you could do additional processing in post prior to making the print or otherwise emitting the image.
-
... (I haven't applied manual sharpening to any of the files; what you see is what I got from the original Lightroom export...
Lightroom files are already sharpened significantly 40/1.0/25/0... are you zeroing this before exporting to ACR?
-
Lightroom files are already sharpened significantly 40/1.0/25/0... are you zeroing this before exporting to ACR?
I'm using ACR's default Detail values: Sharpening 40, Noise Reduction 0, Color Noise Reduction 25. My understanding is that these represent the capture sharpening the ACR developers consider appropriate to apply when the file is demosaiced, and that while the displayed values are invariant what actually happens is specific to the demosaicing algorithm used for the particular type of raw file, but I confess I can't point to an authoritative source for that at the moment. Lightroom also by default applies some capture sharpening (the values you cited), but they were not used in the examples I posted.
The workflow I used was: (1) read the files directly into ACR, (2) perform the "Super Sharpening" in ACR, (3) make some minor tone adjustments in ACR, (4) export a TIFF from ACR; (5) import the TIFF into Lightroom, and (6) make the crops in Lightroom. I did not apply any output sharpening when I exported the crops from Lightroom.
-
Lightroom files are already sharpened significantly 40/1.0/25/0... are you zeroing this before exporting to ACR?
Ah no, not raws.
Sharpening is dependent on ACRs details settings.
-
It seems that the Enhanced option is only there when the raw file is opened directly into PS/ACR. By this I mean that if I open a raw file as a Smart Object from LR then open CR Filter, either from the menu or the layer thumbnail the Enhanced option is missing. Similarly if I open a raw into PS/ACR and then open it into PS as a Smart Object but without Enhanced processing I don't get the Enhanced option if I then take it back into CR filter.
I haven't figured out a workflow for this Enhanced option yet so I don't know if it matters. I just thought I'd point it out in case don't see the option depending on when they open CR. I'm sure there's a reason. I'm on Windows if it matters.
Mike
-
Just played with the new Enhance/Super Resolution. In one word: WOW!
I grabbed an image from the Z 6 and ran it at normal res, then upressed in both PS and OnOne Resize 2020 for comparison. The Super Resolution was substantially better and looked like real pixels than the other two. I don't have Topaz Gigapixel but the few times I've seen it on jpeg and iPhone photos it wasn't too bad.
I tried the SR on some older D800 files. Same impression. Save with my Oly M43 files. To dig deeper, I tried some EM-5 II pix shot with the pixel shift. The first file I tried was an ACR-assembled DNG file from the Pixel Shift, somewhere more than 9,000 by 9,000. That did not seem to work well (got to see the extremes sometimes!). It had weird artifacts from the stitching that looked quite strange. However, when I backed off to a single pixel-shift image (9216x6912), it wasn't quite bad. The finished file came in at 18432x13824) and it's full of detail and still quite sharp, considering it was taken with such a small camera. I wish my printer was large enough to run the full file which now comes in at
When I have more time, I need to print one of these Super Resolution prints. From my Olympus M43, it's now too large to print on my printer, even if I print it at 300 ppi--the image size at that res is just over 61x46. Now if I were to print on canvas, I do believe I could even double this size!
Adobe wowed me with this new feature. I wish I had it last week when I was printing the last project of 30x40s though I think a lot of the new "manufactured" detail would have been lost on the canvas.
Thinking pragmatically, this also may be word-around for lack native-mount longer than 200mm focal lengths for the Nikon Z cameras (though with adapters for the N lenses and the new TC 1.4 and 2.0 on the 70-200, there's reach up to 400mm in Z), without buying a longer lens or simply shoot with the Z 6 "pretend" it's got pixel shifting and more crop potential, provided the image was well crafted from the start!
-
I am getting weird artifacts, in the form of bright and colorful pixels or dots, visible in the Super Resolution preview window.
-
I saw something like that, Slobodan, in an image I was playing with today. I looked a little closer and found a couple more bright, green and red "hot" pixels, maybe three in this one image. I went back to the original raw image and found the same hot pixels there. I think it was probably a combo of a long exposure in a dark church when the sensor did its ring-around. I may have to pay more attention when I use this new tool in real life.
-
I had a look at this feature today. I believe it will be rolled out to LR at some point. Thats good because my workflow is greatly slowed down opening an individual file in ACR directly. Then enhancing it creates a new DNG with the original filename and enhanced appended to it. I tried importing it into LR and while it gets aded top the catalogue LR is unable to view or output it.
Anyway it turned my 70MB A9 file into a 270MB file. IMHO its really very good. I have Topaz Gigapixel and don't like the program at all. It has too much personality for my taste and I really dislike the interface and the workflow. I think this Adobe enhancement is really amazing. I will certainly make use of it once it makes its way into LR which Im sure it will
-
From my first experiments with different kind of photos (cityscape, landscape) :
- natural subjects such as stone, grass, trees get a huge boosts in detail. It's pure magic and giving a complete new life to old images - thanks AI !
- cityscape : some types of architecture gets a boost, some stay the same - overall medium results there. I imagine it's based on pattern recognition and the AI might have difficulty in recognising some building feature. I imagine it can be trained for that
- for better results, the original image has to be sharp in the first place.
IMO this is most significant software addition made in recent times - a complete game changer. In pure resolution terms, would I need to buy a medium format camera if I can get 140MP files from my A7R3 ?
-
According to Eric Chan, it is coming to LRc.
You can save a DNG enhanced from ACR and the next version of LRc will import them.
-
According to Eric Chan, it is coming to LRc.
You can save a DNG enhanced from ACR and the next version of LRc will import them.
Marvelous. This actually impacts on my equipment purchasing decisions in a marginal way. I’m still a bit taken aback by how well this works.
-
Andrew, do you know where it's best to do capture sharpening with Enhanced? It seems there are some options to play with;
- Same setting as normal on original before Enhanced, none(?) after
- None before Enhanced, capture settings after (different?)
- Modified before, modified after
Thx
-
Andrew, do you know where it's best to do capture sharpening with Enhanced? It seems there are some options to play with;
- Same setting as normal on original before Enhanced, none(?) after
- None before Enhanced, capture settings after (different?)
- Modified before, modified after
Thx
First, the order doesn't matter in the final processing; Lightroom Classic and ACR will apply the edits in the order for rendering best applied, not user applied.
If you're doing Capture Sharpening by eye, as we have to, then I'd suggest doing this (or modifying it) after Enhance. Since this would affect what you see.
-
I am getting weird artifacts, in the form of bright and colorful pixels or dots, visible in the Super Resolution preview window.
I get this with iphone dng images, otherwise it works nicely with raw from DSLRs (Nikon in my case)
-
I just updated LR to 10.2 and was disappointed to see that Super Res is not included. The older Enhanced Details is still there.
Also the limitation I mentioned above is still in place, ie ACR in PS will not allow Super Res on a Smart Object opened into PS from LR. You have to open the raw from PS/ACR
-
I am getting weird artifacts, in the form of bright and colorful pixels or dots, visible in the Super Resolution preview window.
+1
Where it works it works great, but at least in my tests with Z6 raw's in man made structures, there are a lot of color artifacts, it's like super resolution creates color aliasing and artifacts from clean sources.
-
Where it works it works great, but at least in my tests with Z6 raw's in man made structures, there are a lot of color artifacts, it's like super resolution creates color aliasing and artifacts from clean sources.
As far as I have been able to determine, that's always a risk when you use this type of technology to enlarge files. The process isn't simply interpolating extra pixels the way traditional enlargement techniques do; it's recreating the image in a larger format based on a machine-learning analysis of various attributes that it detects within the image. If the software "discovers" an attribute that isn't really there from the perspective of a human viewer and then enhances it, you get what the viewer perceives as visual artifacts. That risk can be reduced, as I understand it, by improving the training set of image components that is used in the machine-learning phase.
-
After AI has done its job, you really should use your own CS and scrutinize/retouche the whole photo from top to bottom.
Then you are ready.
-
Can anyone help, please?
I have Photoshop v22.2 installed on my late 2019 MacBook Air with 16GB memory and running Catalina. I have ACR13.1 installed. I understand that I need ACR13.2, but my Adobe update facility does not give me an option to download it. As a result the only enhance option I have is Enhance Details.
Is there away to get ACR13.2 so that I can use Super Resolution?
Thanks in advance for any advice.
Jonathan
-
Usually when the Creative Cloud app doesn't show the update you need to log off CC (not just close it). This forces you to log back on with your Adobe credentials when you re-open it and the updates will show up.
-
Thanks Mike, but when I try to logout of the app, the sign out is greyed and I cannot!
Jonathan
-
Thanks Mike, but when I try to logout of the app, the sign out is greyed and I cannot!
Jonathan
Reboot the computer.
-
Yes reboot. If Sign Out is grey then it means you are not signed in which is why you're not seeing the updates.
-
Thanks Andrew and Mike. Rebooted Macbook and signed in to my Adobe account. Creative Cloud app then stuttered into being, as it seemed what was there was an old version. I got a message that it had been updated and would in future be updated. It then spent a while updating Photoshop, Lightroom Classic and Camera RAW. I now have ACR13.2, and have successfully done a super resolution. It was of a Fuji 26MP RAF image that was some 56MB in size. The super res DNG size is about 360MB (!) and the pixel count is the expected 4x the original.
Thanks for the advice.
Best wishes,
Jonathan
-
I believe it will be rolled out to LR at some point. Thats good because my workflow is greatly slowed down opening an individual file in ACR directly. Then enhancing it creates a new DNG with the original filename and enhanced appended to it. I tried importing it into LR and while it gets aded top the catalogue LR is unable to view or output it.
Try the current (March) release of Lightroom (v. 10.2). It now appears to parse the DNGs enlarged with Adobe Camera Raw without difficulty. Other ACR image enhancements are propagated to LR via an xmp sidecar file. Not seamless, but not overly cumbersome, either.
-
I asked a friend to test the algorithm against a synthetic image to check sharpness improvement on linear details (useful for texts and architectural images). The image was this:
http://guillermoluijk.com/misc/lozoya.png
And the result was quite surprising (I tried to fill some cells in red after the rescaling):
(https://linkpicture.com/q/superresolution.png)
The improvement in detail is fine, but the algorithm also introduces some 'noise' (larger than one level in a 8-bit scale) as an undesired signal vs the expected values, surely as a result of its 'organic' AI nature.
Needless to say the image never went to any lossy format (JPEG), so all this variance comes from ACR's algorithm.
This could become irrelevant for real world photographs where read and photon noise will probably surpass this amount, but it's good to know.
Regards!
-
Try the current (March) release of Lightroom (v. 10.2). It now appears to parse the DNGs enlarged with Adobe Camera Raw without difficulty. Other ACR image enhancements are propagated to LR via an xmp sidecar file. Not seamless, but not overly cumbersome, either.
+1
Jonathan
-
I asked a friend to test the algorithm against a synthetic image to check sharpness improvement on linear details (useful for texts and architectural images). The image was this:
http://guillermoluijk.com/misc/lozoya.png
And the result was quite surprising (I tried to fill some cells in red after the rescaling):
(https://linkpicture.com/q/superresolution.png)
The improvement in detail is fine, but the algorithm also introduces some 'noise' (larger than one level in a 8-bit scale) as an undesired signal vs the expected values, surely as a result of its 'organic' AI nature.
Needless to say the image never went to any lossy format (JPEG), so all this variance comes from ACR's algorithm.
This could become irrelevant for real world photographs where read and photon noise will probably surpass this amount, but it's good to know.
Regards!
Hey, nothing's free in life.
-
Two animated GIF's. First shows the original synthetic image (800%) vs superresolution (400%):
http://guillermoluijk.com/misc/superresolucion.gif
I feel the rescaling enhances detail and sharpness. Another good thing is that it manages to recreate barely visible cells surrounded by close edges in the original image as correct polygonal shapes.
On the downside, noise is already visible. Let's enhance it with a curve:
http://guillermoluijk.com/misc/superresolucioneditada.gif
Noise spreads to areas far from the black edges, where the original image was pure white. Moreover it is coloured noise, while the original image was pure grayscale (Adobe could have taken care of this in case someone rescales monochrome images).
Regards
-
Thanks for posting this topic it will be great help
-
Right or wrong?
If I have an iphone image of 12MP and up res it using super res in Photoshop or Gigapixel, I can have a 48MP image. If I have an image of the same scene taken with, say, a Sony a7r4 and then cropped to 48MP and the same coverage as the other image, which will have more detail? I think it will be the Sony - right or wrong?
Jonathan
-
If I have an iphone image of 12MP and up res it using super res in Photoshop or Gigapixel, I can have a 48MP image. If I have an image of the same scene taken with, say, a Sony a7r4 and then cropped to 48MP and the same coverage as the other image, which will have more detail?
Even assuming you started with a pristine (raw or at least uncompressed) file from the iPhone, you would be comparing a recreated image with an actual one. The AI enlarging tools based on machine-learning analyze the components of an image and reconstitute it in a larger format based on the visual elements they have recognized in the source file. Assuming they do a good job and don't introduce any objectionable artifacts—and in my experience, they often do, but not always—they may be more effective on a given image than enlarging it by interpolating more pixels (e.g., in Photoshop). But I can't imagine a situation when you wouldn't be better off working from the same amount of data collected directly by a camera's light sensor. The value of these tools is to give you more apparent resolution than is available from the camera on those occasions when you really need it.
-
The AI enlarging tools based on machine-learning analyze the components of an image and reconstitute it in a larger format based on the visual elements they have recognized in the source file.
Apropos of which, it's important to select an appropriate "training set" of images which will be fed to the neural network during the recognition phase in order to achieve the desired result. Train the software on images seen through a window covered with raindrops and it will find them in your photo whether they exist there or not. Attached: (1) a source image of the George Washington Bridge; (2) a target image produced by a convolutional neural network trained to find raindrops.
-
Apropos of which, it's important to select an appropriate "training set" of images which will be fed to the neural network during the recognition phase in order to achieve the desired result. Train the software on images seen through a window covered with raindrops and it will find them in your photo whether they exist there or not. Attached: (1) a source image of the George Washington Bridge; (2) a target image produced by a convolutional neural network trained to find raindrops.
I rather like the "bridge in the rain" result. It appears that the neural network has created a new reality.
-
I rather like the "bridge in the rain" result. It appears that the neural network has created a new reality.
This is an example of what I think of as using a convolutional neural network backwards. Instead of having it identify raindrops in an image, use it to recognize raindrop-like visual elements in the source file and make them look more like raindrops in the destination file. Instead of trying to recognize a painting by van Gogh, inject van Gogh-like attributes into the image.
I experimented with this technique on a number of photographs a while back. Some produced interesting results, others didn't. You can see some samples here. (https://www.flickr.com/photos/chriskernpix/albums/72157705322226491)
-
One would like to use super resolution to see more detail in a photo.
indeed at 100% pixel size many details seem improved, yet it goes along with some side effects that are larger than just details
Here two problems i encounter quickly when using the super resolution introduced in 22.3 photoshop.
two versions- 100% pixel size
org- means normal developed and added some basic sharpening in photoshop
better- means superresolution- no sharpening added.
example1
The arrow points at areas with introduced artefacts.
example2
the arrow points at uneven areas with blur/detail that used to be even.
if i use the image smaller on the web, I can see no improvement in detail, but i can see the introduced artefacts.
If i use the image at 300dpi in print I cannot see any difference, but the introduced artefacts.
If i use the image at 100dpi in print I can see the better detail of SR, but also the artefacts.
ergo: in only a small number of cases the SR has an advantage, but the images should be scrutinized to get rid of all the false artefacts.
I am sure that in future versions of photoshop ( or Topaz resize) things will improve.
-
Just used this for the first time on a batch of images from a recent trip to Utah/Death Valley. Pretty damn happy, except for the artifacts creating a "lumpy" sky in a few images.
-
The "Super Resolution" feature is now available in the latest rev of Lightroom (10.3), as well. Seems to work the same as in Photoshop.
-
Super! That just made life easier.
I have to get some new drives. This eats up space quicker than a big American family can wipe out a Golden Corral all you can eat special.
-
Have finally had a real task that needed up res. I was asked to take a high res image of a complex painted map nearly 100 years old and about 6ft x 4 ft 6 inches. It is mounted in a room where one can only get about 8 or 9 ft away. I took a single image and 9 overlapping images in a 3x3 matrix. All were taken in raw with my 26MP Fuji X-T4. I put the single image through Photoshop Super Resolution and the 9 images through Ptgui pro.
My subjective opinion is that Ptgui gave a better result. I am not surprised given that Ptgui did not have to generate pixels. Of course this was a situation that was static.
Jonathan
-
I'm late to the game here because I've been using Gigapixel for over three years and was just reminded of Super Resolution while reviewing back issues of Photoshop User.
Here's my experience with a couple of 20-year-old 1-megapixel images (I like to do "extreme" tests). To my eye, Super Resolution is smoothing out the rough pixelization but not otherwise doing much to these images vs. Gigapixel (?).
I understand they work differently, but I prefer the cleaner results.
-
I prefer to do as much work as possible from raw data.
-
I prefer to do as much work as possible from raw data.
The theory's sound, but in practice I find outputting a full size TIF and then uprezing in PS gives better results than the super resolution option. It's a feature that just not refined enough yet.
Gigapixel offers a marginal improvement over the PS file, but it needs some gamma correction to get back to the correct tonality.
-
Gigapixel offers a marginal improvement over the PS file, but it needs some gamma correction to get back to the correct tonality.
Thanks. Another reason I'll save some money and pass.
-
The theory's sound, but in practice I find outputting a full size TIF and then uprezing in PS gives better results than the super resolution option. It's a feature that just not refined enough yet.
I'm not certain why the ACR and Lightroom versions of "super resolution" would differ. It was my impression they shared the same neural network implementation.
In any event, of your three examples, the one produced by Lightroom super resolution marginally looks the best to me. I see some halos and also softness around the letters in the Gigapixel example. It's as though the program is oversharpening—and then performing aggressive noise reduction that results in blurring the high-contrast boundaries. (Gigapixel Auto Settings option, perhaps?)
I haven't done a lot of head-to-head testing, but on the files I've enlarged with both the Adobe and Topaz products, I've rarely detected any significant difference in 2X enlargements with similar post-enlargement adjustments. Of course, Gigapixel offers options for even greater increases in linear dimensions.
-
I'm not certain why the ACR and Lightroom versions of "super resolution" would differ.
They do not. When on version parity, the two are indentical.
Even without Super Resoution, the ACR engine should provide a better resampling up than Photoshop for a number of reasons, both resampling is a hybrid Bicubic algorithm that interpolates between Bicubic and Bicubic Smoother for upsampling and Bicubic and Bicubic Sharper for downsampling. Lightroom and ACR uses a adaptive bicubic algorithm: the algorithm parameters are chosen automatically based on the relationship between the original image size and the final image size, from raw linear data. These parameters were determined empirically by doing lots of experiments with photos being resampled to common output sizes, such as web-sized images (800 to 1000pixels on the longer dimension), small prints, and big prints.
I can see slight improvement over the same resizing using various Photoshop algorithms. But what I see as visible improvements, others may not agree.
Having parametric edits that just get better as time goes on, and without dealing with large TIFFs (with burned in edits) is yet again why I prefer to do all of this at the raw stage. YMMV.
I agree about the halo's in that 3rd party upsizing product; not attractive to me.
-
It's a feature that just not refined enough yet.
You render a big TIFF, you've baked that cake.
You create a (new) linear DNG with the Super Resolution; your original raw is untouched of course and you're working on not true raw but more raw data but here's the cool thing: Adobe does refine it and other parts of the ACR engine. You still have the original raw, you can resample it using Super Resolution V2 (and V3 etc) and produce a new Linear raw, copy and paste ALL your parametric edits into this new Linear raw. We've seen 5 major process versions in LR/ACR, a 6th will be provided (anything further, I can't discuss).
This again is why I prefer to do as much work as possible from raw data.
-
Having parametric edits that just get better as time goes on, and without dealing with large TIFFs (with burned in edits) is yet again why I prefer to do all of this at the raw stage.
I agree and that's why when I use "super resolution," I don't invoke it until I've adjusted the image to my satisfaction.* I occasionally make some additional minor tweaks to the resulting Linear DNG—perhaps because the neural network has interpreted some of the tonal or color adjustments differently than they looked to me when I was making them on my monitor**—but I don't recall ever feeling the need for more than a couple of post-enlargement modifications. So if at some time in the future I decide to make further changes, I can work from the original raw and spin off another super-resolution enlargement.
———
*I invoke the "fine details" option before I make any other adjustments, however, because it can affect what I do with the manual controls.
**My monitors have a color space roughly equivalent to Adobe RGB, but I presume the "super resolution" function is operating within Lightroom's larger color space.
-
Yes, the ACR engine uses ProPhoto RGB primaries and 1.0 TRC for all processing.
-
You render a big TIFF, you've baked that cake.
Yes, but today I get the best result for today by the route I've outlined and I can then delete if if chose to.
I've still got the raw if the feature is improved at some future time or I can export and use another upsized TIF when I want to, if I deleted it.
-
In any event, of your three examples, the one produced by Lightroom super resolution marginally looks the best to me.
Wow. I doubt many people would agree with you on that.
-
Yes, but today I get the best result for today by the route I've outlined and I can then delete if if chose to.
If you say so (better results). The examples don't speak for themselves as at least two of us have commented.
If you are happy with what you have, be happy that you're happy. :)
Wow. I doubt many people would agree with you on that.
Wow. Two of us here did with respect to your examples.
-
Two of us here did with respect to your examples.
Your comment wasn't exactly clear. But you think that middle example with all the false artifacts and noise is better ? really ?
-
Your comment wasn't exactly clear. But you think that middle example with all the false artifacts and noise is better ? really ?
Two of us agree to disagree with you.
-
ROFL credibility drops another point.
-
ROFL credibility drops another point.
Probably what Chris and certainly I feel about your example. And “workflow”.
To each his own.
-
I've been using Topaz Gigapixel for a couple of years now, and have found it quite useful on occasion (albeit less so since Adobe introduced its "super resolution" option), but it has some quirks I find disconcerting.
First, if you have Gigapixel emit a DNG, what you appear to get is not a traditional Linear DNG but the equivalent of a fully-baked TIFF in a DNG container. Below: the respective file sizes of an RGB raw file (Fuji X-Trans), the 2X enlargement produced by Gigapixel from that raw file, and the 2X enlargement produced by Lightroom from the same raw file:
-rw-r--r-- 1 ck staff 12M Nov 9 18:35 _XT47993.RAF
-rw-r--r-- 1 ck staff 596M Dec 31 16:17 _XT47993-standard-scale-2_00x-gigapixel.dng
-rw-r--r-- 1 ck staff 163M Dec 31 16:21 _XT47993-Enhanced.dng
Although the Adobe super resolution file has been demosaiced (i.e., it's a Linear DNG), it can in some important respects be treated as a raw file—specifically for adjusting white balance and making other color corrections. As far as I can tell—I'm no expert on this stuff, and am prepared to be proved wrong—the Gigapixel DNG is really just a fully-baked TIFF in DNG's clothing.
More irritating from my perspective is that Gigapixel appears to try to "normalize" the tone curve of any raw image it enlarges by stretching out the tonal values so they cover the full range between black and white. That may be what the user wants, but I have not found any way to defeat this behavior and I would rather be able to control the image tones myself: when I use Gigapixel, I just want it to rebuild the image with larger linear pixel dimensions and not perform any default "editing" for me.
-
First, if you have Gigapixel emit a DNG, what you appear to get is not a traditional Linear DNG but the equivalent of a fully-baked TIFF in a DNG container.
If so, that's really dumb and a bit dishonest (giving the impression it's raw simply because of the DNG wrapper). DNG is as I'm sure you know, but many do not, a cousin of TIFF and one can embed a JPEG in it. That's still a JPEG, zero advantage to having it wrapped into a DNG. I wonder how many Gigapixel users think "Wow, a DNG; I've got raw data."
A baked TIFF or JPEG is a baked TIFF or JPEG, no matter how you try to fool people.
At least Adobe is up front about what and when you end up with a Linear DNG.
-
Now that this seems common, do you bother using teleconverters any more? They introduce their own IQ issues, so better to just "Super Res" and leave the teleconverter at home?
-
Now that this seems common, do you bother using teleconverters any more? They introduce their own IQ issues, so better to just "Super Res" and leave the teleconverter at home?
I don't know that there is a single, simple answer to that question. A teleconverter uses an optical approach to project a larger image on the camera sensor. "Super Resolution" uses a computational approach that reconstructs the captured image with larger pixel dimensions. The image quality provided by the teleconverter is affected by the optical limitations of the converter and the particular lens attached to it. The image quality provided by the computational enlargement is limited by the number and variety of the images used to train the neural network to recognize and reconstruct the features in a photograph without producing artifacts. Which approach works best will probably depend on the way the visual aspects of the subject interact with the characteristics of the enlargement method.
-
i have a 2x Nikon converter and with some lenses (70-200) it does not work very well; with others (like the 300pf lens) you get clearly more detail.
Much more than you would ever get with software enhancements, that introduce also a lot of false detail.