Topaz: Edit JPG to Raw Files

[billbane]

Andrew,

Please consider the possibility that you are not clear in your mind about what A.I. means and does, in particular as Topaz is using it. Bart has summarized it well, and perhaps you should reread what he has written.

I was the head of an A.I. Startup some time ago, and the "experts" defined "A.I." as "software that we do not quite understand." This still seems operational.

Some have found that this Topaz product improves old Jpeg files, as I have. Topaz's use of A.I is a first try and could and should become better. A.I. is very different from traditional software, so the traditional "diminishing returns" expectations may be flipped on their heads.

The trial is free; have you tried it? If so, did you try ACR/LR/PS on the result vs using ACR/LR/PS on the original file? When I did, I saw more elbow room in addition to the initial pretty clear improvement. If you have not tried it, please give it a try and let us benefit from your perspectives.

Bill

[digitaldog]

Andrew,

Please consider the possibility that you are not clear in your mind about what A.I. means and does, in particular as Topaz is using it. Bart has summarized it well, and perhaps you should reread what he has written.

I was the head of an A.I. Startup some time ago, and the "experts" defined "A.I." as "software that we do not quite understand." This still seems operational.

Some have found that this Topaz product improves old Jpeg files, as I have. Topaz's use of A.I is a first try and could and should become better. A.I. is very different from traditional software, so the traditional "diminishing returns" expectations may be flipped on their heads.

The trial is free; have you tried it? If so, did you try ACR/LR/PS on the result vs using ACR/LR/PS on the original file? When I did, I saw more elbow room in addition to the initial pretty clear improvement. If you have not tried it, please give it a try and let us benefit from your perspectives.

Bill

I'm quite clear on marketing BS and claims that are impossible. Topaz Labs, A.I. or otherwise have done just so: need to buy beachfront property in New Mexico?

[billbane]

Right on!. Keep on refusing the possibility that your lying eyes might cause you consternation.

[digitaldog]

Right on!. Keep on refusing the possibility that your lying eyes might cause you consternation.

Keep believing in pink unicorns.... And software "we don't quite understand".

[billbane]

Don't know how you can be so certain about this product since you refuse to look. Same approach used by critics of Galileo.

Actually, and fwiw, the unicorns are green after the Topaz conversion, not pink.

[digitaldog]

Don't know how you can be so certain about this product since you refuse to look. Same approach used by critics of Galileo.

Actually, and fwiw, the unicorns are green after the Topaz conversion, not pink.

I know a JPEG cannot be converted to a raw!
Yeah, green unicorns after conversion with Topaz A.I. More marketing fiction. More lies.

[Ray Harrison]

Don't know how you can be so certain about this product since you refuse to look. Same approach used by critics of Galileo.

Actually, and fwiw, the unicorns are green after the Topaz conversion, not pink.

Given a JPEG, Topaz’s application may or may not create a TIFF (or TIFF in a DNG container) that allows you to spit out a better image after some edits. It seems to for some people. That’s great, truly. What it does not do is create a raw file, nor any image that recovers raw-like flexibility (it can’t do that). If they’d called it something like “Magic AI JPEG Enhancer” that’s a closer match.

I definitely get that AI and machine learning can do things to help some images, I work in the ML/AI space, but please, let’s have some reality around what it does. The name of the product simply isn’t an indicator of this. And no, spitting out a DNG doesn’t make it raw.

[billbane]

Ray,
1. Did it allow you to create a better image from the old jpeg?
2. Do you think the others on this blog who say "yes" to this question are deluding themselves?
3. Why are you and others so concerned about using the word "raw"? Have you listened to any TV ads lately that use dull, literal words? Previously I did not give a damn about the product name, but now I must admit that it was brilliant. Look at all the attention that word has given the company. Pure gold. Now maybe they can finance some more R&D and can do even better. They are surely saying, "thank you, thank you, thank you."
Bill

[digitaldog]

How exactly does placing a JPEG in a DNG, or saving it as a TIFF give more editing headroom?
Yes, it certainly doesn't become raw data.
Yes, a converted TIFF or DNG after a subsequent edit, will not suffer additional JPEG data loss as if the edits were applied and saved again as a JPEG. More data and overhead otherwise?

[digitaldog]

Ray,
1. Did it allow you to create a better image from the old jpeg?
2. Do you think the others on this blog who say "yes" to this question are deluding themselves?
3. Why are you and others so concerned about using the word "raw"?

Because it's a friggin lie and utter BS, that's why. Like my offer to sell you beachfront property in NM. But a group of suckers are born every minute.
This of course has nothing to do with the ability of this software or any image editing software to make an image that require improvement to be improved. I don’t know if you and others defending this products nonsense are purposely trying not to understand this, or if you all are really struggling with it.

[Ray Harrison]

Ray,
1. Did it allow you to create a better image from the old jpeg?
2. Do you think the others on this blog who say "yes" to this question are deluding themselves?
3. Why are you and others so concerned about using the word "raw"? Have you listened to any TV ads lately that use dull, literal words? Previously I did not give a damn about the product name, but now I must admit that it was brilliant. Look at all the attention that word has given the company. Pure gold. Now maybe they can finance some more R&D and can do even better. They are surely saying, "thank you, thank you, thank you."
Bill

1.) No
2.) They had different use cases, I’m assuming, plus I never indicated that someone couldn’t get something they liked better. Great! That’s what tools are for.
3.) Raw means something, spitting out a tiff or a dng isn't that something. The end result of the Topaz application doesn’t give me the flexibility of a raw file, or behave in any way like one. And yes, they’ve made a silly, possibly controversial, and certainly inaccurate name. I’m personally not going to help “finance some more R&D”. It definitely sounds like you are. I’m super excited for you that it fixes all your JPEGs!

If I was selling a machine, let’s say “Ray’s Lead to Gold Converter”, and for the sake of argument , it didn't actually do that, maybe it just spray painted your lead a gold color instead of actually changing elements. And let’s say I wasn’t selling it as a joke. I would hope people would call me out on it, and that I couldn’t just get away with saying “Hey, if I used dull, literal words, nobody would buy it so I used GOLD instead!”

[billbane]

Andrew,

Bart can explain better all of this (and has), but I continue to think you are trying to force this product into some past analysis and insights you have had. Try with a fresh sheet of paper. Forget terms like jpeg, raw, Dng, and Tiff as you have understood them.

Maybe I can get down and dirty and such a description might help understand what Topaz is attempting.....

First, what I understand Topaz to be doing might be thought of as like PS's "content aware fill", which I think is also supposedly being allowed/enhanced by some of their own "A.I.". How does it work? PS "looks at" the pixels that are selected as acceptable replacement pixels. Base on that selection, I think (but do not know) that it then generalizes about the default selected pixel set and maps those acceptable pixels via a transformation matrix (which is what I think the A.I. generates) into the pixel space where the selected fill area has been selected. Sometimes it does it well and sometimes terribly. The new version from Adobe lets humans see which pixels the algorithm is allowed to use as acceptable substitute pixels and pixel patterns. When the user changes this acceptable area, think of that as "training" PS to use a different transformation matrix. Maybe the result is better, maybe not, maybe perfect, and the result will depend on the complexity of the "source" pixels and the target pixel space. Should we reject this and confine ourselves to only using a pure cloning mechanism? The new pixels are not "real". Do these improve images contain pink unicorns?

Again shooting in the dark, imagine a x by y pixel matrix, aka the jpeg you see on your screen. Imagine banding and other jpeg artifacts one commonly sees. What the Topaz paper says (I take some liberties in the interest of brvity and possibly clarity) is that they took many paired raw/jpeg examples, and did some "A.I" on them (think of it as Excel regression analysis), pixel coordinate by pixel coordinate for the entire set of images. Since each pixel coordinate has many, many attributes like tone, color, and others, the software goes to the first pixel and creates a data matrix of differences between the raw version pixel and the jpeg pixel. Now do this for every set of pixels on that particular image. Now find the next x thousand images and do the same. Now you have a boatload of data equaling the number of images by the number of pixels times the set of data differences. Now, "regress" all of this and boil it down into a very complex mathematical polynomial. Now invert this polynomial equation and try it on some jpeg image not in the analyzed data set. If it is not ok, do all of this again but twiddle with the knobs to get a different polynomial. etc, etc. It is certainly more complicated than this, but on the other hand, it is just trying to find repetitive patterns in the many Raw to jpeg conversions it analyzed. Each of you know how to describe what banding in the sky looks like, so all these equations do is try to "undo" that causation via mathematics.

Clearly, this has nothing to do with the pattern of data in a camera raw data set. Maybe, if it is good, it will create a pattern of data that, when stored and opened as, say, a DNG file format, it creates a pattern of pixels that better approximates what the original raw data set represented than the jpeg image does. If, in the process, it creates more gradations, say, between this tone and that, it gives you and me, the PS mechanics, more choices of, say, intermediate tones to tune to your taste. If so that would be or could be more "headroom."

Content aware fill, Topaz J2r. Use'um if you like, forget'um if you wish. Why their names matter in "image processing" continues to escape me. My recollection is that many "real artists" laughed at users of Photoshop as fake art and laughable. They are less sure now. Most importantly, A.I. is a basis to improve image processing, and it will get better and better, and no matter how well content aware fill and J2R perform now, odds are they will become better and better. Objecting to this as a possibility seems very odd to me.

All this literalism about the word, "Raw" recalls Alice in Wonderland when the Humpty Dumpty says about words that they meant whatever he said they meant. This seems even more odd than improving imaging processing. Nobody can patent word meaning. Businesses and communities routinely change the meanings of words and add meanings. Raging against it will not change anything.

Bill

PS. "Sep 10, 2018 - Everyone went nuts for Adobe's “content-aware fill” in Photoshop when it ... essentially an AI-powered clone stamp that intelligently brought in ..." from https://techcrunch.com/2018/09/10/adobe-supercharges-photoshops-content-aware-fill-so-you-have-more-options-fewer-ai-fails/

[billbane]

Ray,

Sorry to tell you, but each time you buy a product, you "fund their R&D".

Topaz is a pimple competing with Goliath Adobe. Compared to Topaz's loose use of the word "raw", Adobe's broken promises makes them look Saint like.

You pull for the big guys, or in your case your desired profitless companies, but I pull for the inventors who try to invent something that will help me.

Bill

[Ray Harrison]

Ray,

Sorry to tell you, but each time you buy a product, you "fund their R&D".

Topaz is a pimple competing with Goliath Adobe. Compared to Topaz's loose use of the word "raw", Adobe's broken promises makes them look Saint like.

You pull for the big guys, or in your case your desired profitless companies, but I pull for the inventors who try to invent something that will help me.

Bill

Bill, sorry but I’m not specifically sure what point you are trying to make, or even where Adobe comes into the picture. I like inventors too. Most people here do. I have topaz products and products from a lot of other smaller companies. But I also think words have meaning and it’s how we apes communicate. Yes, words evolve, I get that. Right now though, raw means something specific, as does dng, tiff, jpeg. Topaz has created a ML-based jpeg to tiff image processor based on interesting training sets. It’s certainly clever image processing tech, it may help people who have a bunch of JPEGs in certain cases, and as you point out will get better over time. All cool. If a jpeg to tiff tool is valuable in the market place, it will certainly sell well. If it helps you to be fuzzy with your definition of raw, knock yourself out. I disagree, though.

I’ve spent more time than necessary already (and wasted a perfectly great Colorado evening 🙂) on a product that I don’t need (no JPEG-only images) and certainly am not purchasing after the trial. I’m glad it’s full of awesomeness for you.

[Doug Gray]

RAW files are universally understood in the photo industry to refer to unprocessed digital data from the camera's sensor together with metadata containing info such as lens used, exposure settings, etc.

A RAW image, being unprocessed lets one do many things such as:

1. Develop a scene referred image. This does not impose a tone curve and is used for making good copies of existing art. When you take a typical, jpeg, output referred picture of a picture and print it the same size it will have more contrast and more saturated colors in the midrange with less contrast and more desaturation in the dark and light extremes. It will be a very poor copy. OTOH, if you use the same settings to develop a RAW file of a normal scene the result will not be very pleasing.

2. Make highly accurate copies of black and white printed documents. Doing this requires bypassing the CFA  and scaling the RGGB channels individually. dcraw, for instance, has an option to do just that.

It may well do a great job detecting and removing jpeg artifacts and make a nicely improved 16 bit tiff file (or one wrapped in a DNG) but it can't create RAW files that can be developed as above. The information to do so just doesn't exist in the jpeg. That cake has been baked.

Call it whatever they like but don't abuse the language. Currently RAW has a clear meaning to photographers. Let's keep it that way.

[albytastic]

Frankly I don't care about the name - my question is "Does it work?" and it does!
Amazingly so - and can be used to produce virtually noise free images in the process.
They use TIFFs because what you get from the JPEGs is a 16 bit image so obviously cannot be a JPEG.
I haven't tried to check this yet but I know that images I've produced by it certainly seem smoother when editing.
But if all you can do is argue that it can't work because it's called raw then that is nonsense, since it does work.
And it certainly seems to recover some lost highlights.
I've put some more comparison photos on my Flickr site so just go there and see for yourselves.
Or better still, actually try it out.
After all, since you get a free 30 day trial you have nothing to lose - except perhaps, your prejudices.

[Ray Harrison]

Frankly I don't care about the name - my question is "Does it work?" and it does!
Amazingly so - and can be used to produce virtually noise free images in the process.
They use TIFFs because what you get from the JPEGs is a 16 bit image so obviously cannot be a JPEG.
I haven't tried to check this yet but I know that images I've produced by it certainly seem smoother when editing.
But if all you can do is argue that it can't work because it's called raw then that is nonsense, since it does work.
And it certainly seems to recover some lost highlights.
I've put some more comparison photos on my Flickr site so just go there and see for yourselves.
Or better still, actually try it out.
After all, since you get a free 30 day trial you have nothing to lose - except perhaps, your prejudices.

No one anywhere is saying the tool doesn’t potentially do something to make certain images better. It’s clear that it is indeed an image enhancing tool that uses machine learning and that people are finding value. While I’m not the target audience of the tool as I don’t have a lot of JPEG-only images, I have tried it with some difficult images, seeing if it would recover some shadow detail from underexposed somewhat noisy files. I’d expect to be able to from a raw file. And it didn’t perform to my expectations. Not even close. To be up front, I didn’t really expect it to because I know enough about how JPEGs are produced and what gets lost in translation. My expectations, skeptical as they were, were driven by the name, which you and others are arguing is meaningless.

I’m not writing more on this subject as it has clearly devolved into absurdity, but I’m glad the tool does what you need!

[Bart_van_der_Wolf]

All that JPEG to RAW aims to achieve is, to create a higher quality source file to base one's editing on.

That source file is not a Raw in the traditional sense, but it offers much of the flexibility of the original Raw file that produced the JPEG.

The result is not perfect, but it does eliminate JPEG compression artifacts, removes noise, adds some of the lost/missing detail, and it improves the robustness of the 'New original'. There is room for improvement, e.g. when a training model is added that produces less noise reduction and sharpening.

Case in point, I've taken a frame from a movie/video (from Blade Runner 2049), and output that as a DNG which was then processed by Capture One Pro. The resulting 'Raw conversion' produced a Wider Gamut TIFF, which was dumbed down to a maximum-quality sRGB JPEG attachment.

The IMHO too aggressive noise reduction caused some (avoidable) shadow detail loss, the overly aggressive 'sharpening' had to be reigned-in a lot with a -60 structure adjustment in CO12, and the default color balance was improved too much because it removed the original colorcast.

So there are several aspects that could be improved (and probably will over time, just like already happened to A.I Gigapixel), but except for the noise reduction, the more robust file allows to post-process it in any way one wants. A brittle JPEG would not allow the level of postprocessing that the DNG allows.

As a side note, I'm reading a lot about people who are upgrading (fragments of) their Videos by using A.I. Gigapixel on the individual JPEG frames and then converting those back to Video. So while not intended for that purpose, a new use has already been found, and the batch processing capability comes in handy when 10s of thousands of images need to be processed. I've also attached a small crop from a 6x upscaled image from the same video. That improved resolution (smaller detail than 6 pixels), but also shows the very narrow DOF in movie frames (which is compensated for by viewing movement from a distance).

JPEG to RAW's potential lies in removing the typical JPEG artifacts (block artifacts and loss of chrominance detail) and postprocessing brittleness, so maybe video frames are of relatively too good quality to show the benefits. But average quality phone/camera JPEGs do already benefit a lot, and things will only get better.

Cheers,
Bart

[Slobodan Blagojevic]

Simple: report them to the Federal Trade Comission for false advertising.

https://en.m.wikipedia.org/wiki/False_advertising

[Ray Harrison]

Bart,
Many thanks for an actual level headed and honestly useful overview!

Thanks,
Ray