-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
How is the Demosaicing in On1 for Mac?
- Thread starter Glen Barrington
- Start date
Marathonianbull
Leading Member
Most people don't, that's why they end up satisfied with whatever they paid for following agressive marketing schemes (Skylum
)
tbcass
Forum Pro
How do you tell the difference?Most people don't, that's why they end up satisfied with whatever they paid for following aggressive marketing schemes (Skylum
Marathonianbull
Leading Member
How to tell the difference?
Simple when you compared several results of just-opened raw files then saved as is into jpeg. In other words...
In photography editing, demosaicing refers to the process of converting the raw data from a camera's image sensor (which usually has a color filter array, often a Bayer pattern) into a full-color image. The sensor captures data in red, green, and blue channels, but each pixel only records one color (due to the filter array), so the software must interpolate the missing colors for each pixel. The quality of this process can vary, and there are distinct differences between good and bad demosaicing.
Good Demosaicing:
Have a nice day Tom, but by the way what is your favourite photo editing software right now? In my case, it used to be LR but now as mentioned earlier I would say Exposure X7 (although it hasn't been renewed for a while; am still waiting to try out SilkyPix 12 when a MacOS version comes out!)
Simple when you compared several results of just-opened raw files then saved as is into jpeg. In other words...
In photography editing, demosaicing refers to the process of converting the raw data from a camera's image sensor (which usually has a color filter array, often a Bayer pattern) into a full-color image. The sensor captures data in red, green, and blue channels, but each pixel only records one color (due to the filter array), so the software must interpolate the missing colors for each pixel. The quality of this process can vary, and there are distinct differences between good and bad demosaicing.
Good Demosaicing:
- Accurate Color Reproduction: Good demosaicing produces natural, accurate colors. It correctly interprets the raw sensor data and minimizes color artifacts like color bleeding or incorrect hues. You should see smooth transitions and accurate color gradients in your image, especially in areas of subtle color differences.
- Detail Preservation: High-quality demosaicing algorithms preserve fine details, particularly in areas of high contrast and texture. Good demosaicing reduces the appearance of aliasing artifacts (like moiré or jagged lines) and creates a smooth image with minimal visual noise.
- Minimization of Artifacts: Good algorithms minimize unwanted artifacts such as color fringing (chromatic aberration) or zigzag patterns (false color patterns). These artifacts can appear as colored halos around edges or over-saturated color patches that look unnatural.
- Noise Reduction: High-quality demosaicing reduces the appearance of sensor noise, especially in low-light conditions, while still maintaining sharpness. It strikes a good balance between smoothing out noise and keeping important fine details intact.
- Efficient Edge Handling: Good demosaicing algorithms handle edges in the image intelligently, without introducing harsh transitions or artifacts. The result is an image with well-defined edges that still look natural and crisp.
- Dynamic Range Preservation: Good demosaicing algorithms do not over-smooth the image or underexpose highlights or shadows, preserving the dynamic range of the raw file. This allows for better post-processing flexibility, particularly when recovering highlight and shadow details.
- Color Artifacts: One of the hallmarks of poor demosaicing is the appearance of color artifacts, such as color fringing, incorrect color saturation, or odd color casts in areas that should be neutral. These artifacts can make the image look unnatural and unpleasant.
- Loss of Detail: Bad demosaicing tends to blur fine details in the image, especially in textures, fine lines, and small objects. It may result in images that appear overly soft or lack sharpness where it should be maintained.
- Moiré Patterns: Poor algorithms often fail to handle patterns like fabric textures or grids and may produce moiré patterns—those wavy or rainbow-colored patterns that don't exist in the original scene. This is a classic sign of a demosaicing failure.
- Blockiness or Zipper Artifacts: A common issue in bad demosaicing is the appearance of blockiness or zipper-like artifacts along straight lines or edges. These are often the result of poor interpolation between pixel colors.
- Haloing and Ringing: Haloing or ringing artifacts, often around high-contrast edges, are a sign of bad demosaicing. These can appear as unnatural, halo-like glows or rings around edges that were not present in the scene.
- Noise Amplification: Poor demosaicing might amplify sensor noise in the image, especially in low-light conditions. This means that post-processing might become more difficult, as you would need to deal with excessive noise while trying to retain detail.
- Incorrect White Balance: Poor demosaicing can result in significant white balance errors, where the colors look off or unnatural. This is especially noticeable when editing raw files, as the raw data can have a different temperature than expected, and bad algorithms can exacerbate the problem.
- Good demosaicing preserves natural colors, detail, and sharpness while minimizing artifacts and noise, which results in a high-quality, flexible image for further editing.
- Bad demosaicing introduces artifacts like color fringing, moiré, noise, and loss of detail, which leads to an unnatural and low-quality final image.
Have a nice day Tom, but by the way what is your favourite photo editing software right now? In my case, it used to be LR but now as mentioned earlier I would say Exposure X7 (although it hasn't been renewed for a while; am still waiting to try out SilkyPix 12 when a MacOS version comes out!)
What difference in demoasicing do you see in the comparison I posted here a month ago?
https://www.dpreview.com/forums/thread/4790679#forum-post-68098739
All I see is a difference in color rendition. Is that what you consider bad? I could have adjusted the results to produce similar colors, but that factor is under the user's control, so I don't see it as an issue. If demosaicing destroys detail or adds artifacts, that's an issue.
I asked another poster at that time to show an example supporting his statement about good and bad, but that never happened. Nobody other than myself has posted examples supporting anything said so far about ON1 demosaicing. I'm asking you again to post an example supporting the one you made here:
https://www.dpreview.com/forums/thread/4790679?page=2#forum-post-68147077
Also, what's the explanation for the idea that demosaicing under Mac is different from demosaicing under Windows? Is it because the Mac operating system does things during demosaicing that that Windows doesn't do? If so, what are those things?
Last edited:
Glen Barrington
Forum Pro
As an m43s user, I have long held that some software is better for m43s than other software. I have long held that ACDSee Windows products are about as good for m43s ORFs as one can get. I think it is at least as good as OM Workspace, but with a pro grade DAM. This might be my subconscious evaluation of the results.
Having switched from Windows to Mac, I'm struggling to find software I'm happy with. I don't really like Apple photos for example yet many people in the Mac community seem to like it. I wonder if Demosaicing quality can vary by camera brand. The ACDSee for Mac products isn't quite "right" in my mind. Good, but not quite as good as the Windows products.
I'm currently using On1 primarily because it's cheap and highly regarded. But I am uncomfortable with it. (Beyond just the weird user interface!). I think your approach might help clarify what I am looking for.
As I asked in a post above yours: What's the explanation for the idea that demosaicing under Mac is different from demosaicing under Windows? Is it because the Mac operating system does things during demosaicing that that Windows doesn't do? If so, what are those things?
Why would ACDSee under Mac use different algorithms from ACDSee under Windows?
The other poster's 'approach' lumps numerous actions and effects under the heading of demosaicing. The more correct term for that collection of actions would be processing.
I'm thinking that people are not in agreement as to what demosacing actually means, and some are using it to refer to almost everything an app does in the chain of actions that result in a finished image. That is not the way I use the term.
Last edited:
Marathonianbull
Leading Member
ishwanu, there are no scientific answers to your / our desperate quandary, and his or my opinion based on personal experiences is as good as yours! Having said that, I really enjoyed looking at your pictures comparing ON1 Photo RAW and DxO PhotoLab. Obviously, some aspects seem different (reds from ON1 appear much stronger, for instance). Which colours overall look closer to your subject? Which picture can preserve more details in highlights and shadows when playing with the respective tonal response curve? We could spend months debating uselessly. My point is: I own several past as well as current versions of several photo editing software, including ON1, but always end up preferring one or two (???) over the others (???). Enjoy your art ; )
Glen Barrington
Forum Pro
Clearly, I don't have a strong personal definition of what demosaicing is. But my 40 plus years of photography experience, I think, allows me to have a strong opinion of what makes a good starting point for my m43s photography. And I have noticed discrepancies.
It may well be that On1 is as good as is possible on the Mac, I don't know. But certainly this evaluation method can't hurt.
I can't argue this topic from my technical knowledge, but I can use whatever advice that comes my way and see if it resolves my questions. This I am willing to do.
It may well be that On1 is as good as is possible on the Mac, I don't know. But certainly this evaluation method can't hurt.
I can't argue this topic from my technical knowledge, but I can use whatever advice that comes my way and see if it resolves my questions. This I am willing to do.
tbcass
Forum Pro
I use DXO Photo Lab 8.
Marathonianbull
Leading Member
I have always liked the demosaicing (not necessarily further processing actions) of ACDSee for Mac (so far w/ version 5, 6, 7, 8, 9 that is), and it may be wrong info but apparently it simply uses the one included by Apple in its OS! As do others (including Pixelmator Pro; excluding ON1, Camera Raw in Adobe products, SilkyPix, RRP 64, and some others forcing their own).
Last edited:
Reilly Diefenbach
Forum Pro
It's easy enough to compare the downloaded raws from DPR Sample Images for your camera.
DPR LRC Conversion
DPR
If there are few artifacts or excessive moire and false color, you know you're on the right track. Trying various (high) ISOs is a good idea as well.
Marathonianbull
Leading Member
That is my next big purchase, waiting for a sale if ever...
tbcass
Forum Pro
Black Friday next November.
I have read that MacOS provides some kind of built-in actions for image processing, but was never clear on what they are, or if they can be bypassed in order to use corresponding actions contained in third-party software instead. If it's not possible to bypass the internal actions, that seems undesirable.
I addressed that in my comments. There are certainly differences in color, but attributing them to demosaicing is IMO misleading. Much more happens after demosacing in PhotoLab, for sure, even when all other adjustments are turned off. PhotoLab applies some kind of 'under the hood' tone curve and doesn't even offer a linear profile (I used one provided by a third party). ON1 Photo RAW does offer a linear profile, though I can't say if either is doing what it should be doing.
That has little if anything to do with demosaicing. It's one of the many things determined by the way later adjustments work.
No, we should quickly be able to clarify what is meant by demosaicing as opposed to other actions applied by software.
I fully understand having a preference for the results that can be obtained with one product as opposed to another, but I don't understand the idea of attributing so many different aspects of that to the single action of demosaicing, which is just one step: CFA interpolation.
https://en.wikipedia.org/wiki/Demosaicing
Last edited:
I agree that moire and false color anomalies can often be attributable to demosaicing algorithms. But I think some posters are attributing things to demosaicing that shouldn't be attributed there ... and the lack of comparative examples doesn't help.
Last edited:
Glen Barrington
Forum Pro
Maybe I need to re-evaluate the ACDSee for Mac product. I really like the DAM and if most 3rd party software is using what Apple provides, maybe I should select the software based on other factors. Certainly, I'm seeing really good work from ALL sorts of software.
Of course you can bypass that internal offering, it is just very enticing not to have to worry about the demosaicing and concentrate on the editing side of things (and even then the engine built into the operating system makes some things really easy to work with and they are highly optimized to make use of the available hardware). DxO, Adobe, On1, Capture One do use their own algorithms - because they do things on that level that the Apple provided algorithms don't offer. AFAIK ACDSee (possibly), Apple Photos and RawPower/Nitro and a whole lot of others do use the Apple provided engine - it has a lot of potential but it is quite cryptic to configure - see RawPower and the documents that the software developer provided along side it - he is one of the former Aperture developers and is uniquely capable of leveraging the full power of the Apple provided core engine.
Some products use the open source libraw (it may be at the core of On1 as well) such as Luminar Neo and of course the open source RawTherapee and Darktable. The benefit of the libraw use is the plethora of choices in algorithms available for demosaicing, the biggest downfall of these is the plethora of choices in algorithms available for demosaicing - the problem is, you the user are supposed to know which one is the best for that individual photo you are going to work on. If you choose the right one (assuming that you are given the choice, which is not a given), one specific property of the converted image (noise, color fidelity, contrast, initial sharpness without sharpening, dynamic range, roll off to the highlights) might be outright the best of the best available but to the detriment of the other properties. There may be none of the algorithms that truly fits your expectations. The commercial demosaicing algorithms are tuned to be universally suited for every possible image and they may leave a hair of sharpness, a hair of contrast, a hair of noise, when compared to the best of the best in any of those properties but they will be close second without big sacrifices. This fine tuning is missing from all libraw users
Last edited:
I would say that only a very small minority of programs is using the Apple provided solution - unless it is an Apple only solution, an even smaller group of those that do are able to leverage the full potential of the built in engine (if I am honest, only Rawpower and Nitro do so - because they are developed by a former Aperture developer and this Apple inbuilt engine was uniquely geared to be useful to the Aperture developers).
Similar threads
Keyboard shortcuts
- f
- Forum