OMDS OM-1 workflow

[BobT]

Lightroom Version Number

11.2

Operating System

1. Windows 11

A question to OM-1 users. Up till now, I've been very happy importing OLympus ORF files to LrC, converting to DNG and doing all PP there. Things have changed with the OM-1 camera. I notice that in order to get the best out of the ORFs, some are importing direct to OM Workstation then exporting to LrC as 16bit TIFFs for further processing. I really don't want to have a two stage process nor the larger TIFF files. Is there a better way?

 

[Ken @ Canadian Rockies]

Side note: I have always used my 16bit TIFF scans. And when I get around to them with Topaz Denoise, they clean up nice. Then I save as an 8bit. Yes, they can be huge files.

 

[KarstenG]

I notice that in order to get the best out of the ORFs, some are importing direct to OM Workstation then exporting to LrC as 16bit TIFFs for further processing.

IMO, the main reason for this is to make use of the advanced denoise capabilities in OM Workspace, to achieve similarly clean results as with OM-1 OOC JPGs.

OMDS have marketed the OM-1 to have "35mm equivalent noise performance"; in nearly every product intro and review it has been stressed that even at very high ISO numbers the (JPG) images have no or very little noise. Unfortunately, this is not coming directly from the new sensor but it is accomplished by software. A new denoise algorithm which is running on the TruePic X (inside the OM-1), as well as part of OM Workspace.

Currently, the only way to achieve clean images from noisy RAW/ORF files is to go through OM Workspace.
Capture 1 v22 and LR 11.2 can process OM-1 ORF, but at least for LR the support is not yet complete as per Adobe.
Personally, I doubt that LR with its current denoise technology will be able to accomplish similarly clean results as OM workspace can do.

I also own the OM-1, and use it for bird photography where you often have to deal with higher ISO but still want to retain as much detail as possible. I'm following the same workflow as with the E-M1 III - importing the ORFs into LR, and pass them on (as TIFFs) to Topaz Denoise to clean them up if needed. Which gives even better results than what OM Workspace can do

 

[BobT]

IMO, the main reason for this is to make use of the advanced denoise capabilities in OM Workspace, to achieve similarly clean results as with OM-1 OOC JPGs.

OMDS have marketed the OM-1 to have "35mm equivalent noise performance"; in nearly every product intro and review it has been stressed that even at very high ISO numbers the (JPG) images have no or very little noise. Unfortunately, this is not coming directly from the new sensor but it is accomplished by software. A new denoise algorithm which is running on the TruePic X (inside the OM-1), as well as part of OM Workspace.

Currently, the only way to achieve clean images from noisy RAW/ORF files is to go through OM Workspace.
Capture 1 v22 and LR 11.2 can process OM-1 ORF, but at least for LR the support is not yet complete as per Adobe.
Personally, I doubt that LR with its current denoise technology will be able to accomplish similarly clean results as OM workspace can do.

I also own the OM-1, and use it for bird photography where you often have to deal with higher ISO but still want to retain as much detail as possible. I'm following the same workflow as with the E-M1 III - importing the ORFs into LR, and pass them on (as TIFFs) to Topaz Denoise to clean them up if needed. Which gives even better results than what OM Workspace can do

I'm seeing the same sort of thing. Importing a standard 20MP ORF into Workspace then exporting to LR as a TIFF without further treatment, produces a result similar to applying LR's Raw Detail enhancement but with slightly better colour. Doing the same with thing using the camera's Hand Held High Res mode produces way superior results to applying just LR's Raw Detail. Adobe has a bit of catch to do. My HHHR ORF file was 41MB but unfortunately, exported as a TIFF, it became 293MB so needs to be used sparingly.
OM Workspace has still a further tick up it's sleave with its AI Noise Reduction. Unfortunately, this is out of my reach because I don't have the required Nvidia GPU. This is where LR has the advantage. LR's Super Resolution + Raw Detail works happily with the integrated graphics on the Intel i7 CPU.

 

[KarstenG]

OM Workspace has still a further tick up it's sleave with its AI Noise Reduction.

In fact OM Workspace has two different options for noise reduction. AI NR which requires Nvidia GPU is pretty powerful (and comes at zero cost to those which are not willing to invest into Topaz or DXO), but even the "standard" NR that doesn't requires special HW is very good and a lot better than what LR can do.

To be honest, I have never worked with RAW Detail but fully rely on Topaz and DXO if I need special treatment on an image.

 

[BobT]

In terms of denoise and sharpness, of late Lr is being challenges by DxO and Topaz but now it's being trounced by a freebie, albeit a brand specific freebie. Perhaps Adobe has been caught napping. I sure hope the Adobe minions are busily working on this. At the moment, it's one brand that has dipped its toe into the computational photography pond but others cannot be far behind.

 

[Denis de Gannes]

A question to OM-1 users. Up till now, I've been very happy importing OLympus ORF files to LrC, converting to DNG and doing all PP there. Things have changed with the OM-1 camera. I notice that in order to get the best out of the ORFs, some are importing direct to OM Workstation then exporting to LrC as 16bit TIFFs for further processing. I really don't want to have a two stage process nor the larger TIFF files. Is there a better way?

This is certainly a significant step forward for OM Workspace, however it applies only to a few recent camera models. The ORF raw images from my OM-D E-M1 do not benefit from the new features. However I do benefit from the recent Enhance and Super- Resolution features in LrC.

 

[BobT]

In fact OM Workspace has two different options for noise reduction. AI NR which requires Nvidia GPU is pretty powerful (and comes at zero cost to those which are not willing to invest into Topaz or DXO), but even the "standard" NR that doesn't requires special HW is very good and a lot better than what LR can do.

To be honest, I have never worked with RAW Detail but fully rely on Topaz and DXO if I need special treatment on an image.

Having had more time to play with OM Workspace now, I agree with you that it does a better job than I can achieve with LrC, marginally better but still better. LrC seems to leave more sharpening artefacts. The curious thing is that WS does this without intervention. I import an ORF, export a TIFF and all done. All there is for me to do in LrC is finish off and catalogue.
As you know WS offers their AI Noise Reduction as a downloadable add-on. I don't have the GPU required to run it. I'm thinking of getting one but we are talking of hundreds of dollars. Have you experimented with AI NR? I'd like to know your thoughts before making a commitment. I expect there would be little gain in low ISO well exposed shots but a greater benefit high ISO poorly lit shots. Is this so?

 

[KarstenG]

Have you experimented with AI NR? I'd like to know your thoughts before making a commitment. I expect there would be little gain in low ISO well exposed shots but a greater benefit high ISO poorly lit shots. Is this so?

I'm sorry I can't give you a fact-based answer, Bob. I think I have never even tried the AI add-on so far (because I'm not using WS at all, except for de-fishing). In general I would assume the same as you - no benefit in low ISO.

If it helps and nobody else here can share her/his experience with WS noise reduction standard vs AI, I could produce a side-by-side comparison for a low ISO (eg ISO 800) and a high ISO shot (eg ISO 6400).

 

[KarstenG]

Bob -

I have prepared a set of files for a poorly lit ISO 6400 butterfly shot:

Filename	Content
0_IMG_RAW.jpg	LR export of original file. No NR, no sharpening.
1_IMG_WS_LOW.jpg	OMWS standard settings and NR "LOW"
2_IMG_WS_STD.jpg	OMWS standard settings and NR "Standard"
3_IMG_WS_AI_LOW.jpg	OMWS standard settings and AI NR "LOW"
4_IMG_WS_AI_STD.jpg	OMWS standard settings and AI NR "Standard"
5_IMG_Topaz.jpg	Topaz Denoise AI: denoised and sharpened using "RAW" model.

You can download the files from my dropbox to inspect them.

To me, for this example, AI noise reduction in OM Workspace yields slightly better results than the standard method. In full screen, you can still figure that there is noise. If you want my opinion - I believe this is not an improvement, which justifies investing into new GPU HW.

The big step forward which you might be hoping for comes into play only with Topaz Denoise. No visible noise in full screen view, and significantly better preservation of details (use LR's comparison view to examine the difference in details on the wings between files 4 and 5).

Hope that helps to gain clarity.

 

[BobT]

Karsten,
Thanks for that. It's most helpful. It answers my question.
A recent observation. For well lit subjects, I've found WS noise reduction slightly superior to LrC. However, for poorly lit misty subjects, such as distant bushland landscapes, it doesn't distinguish well between noise an detail and considerable detail is lost compared to the same amount of noise reduction in LrC.
On another tack, I'm wondering how LrC deals with HR ORFs. Would it recognise them as pixel shift shots and apply the appropriate demosaicing algorithm? I expect they would have to be processed through WS first? I'm thinking, there are probably generic pixel shift algorithms around but OMDS may well use their own tweeked version. Any thoughts?

 

[KarstenG]

Happy to hear this, Bob!

it doesn't distinguish well between noise an detail and considerable detail is lost compared to the same amount of noise reduction in LrC.

This is a typical challenge for any NR software. In most tools, you have two adjusting screws to balance NR vs details preservation. In WS, you can just control the NR level, plus there's a binary priority setting in the AI-based method. I believe this is not sufficient for ambitioned users, to get out the best of their images.

I'm thinking, there are probably generic pixel shift algorithms around but OMDS may well use their own tweeked version. Any thoughts?

Regarding capturing and processing of HiRes files, there is an interesting thread in DPreview forum.
If you don't want go through the entire discussion, check the first post and this one.
Takeaway: it's very difficult if not even impossible to produce similarly sharp results in ACR compared to WS. The suggestion is to use WS as a HiRes ORF preprocessor. Perfectionists may even think about blending WS and ACR outputs to get the best of both worlds...

As far as I'm concerned, I have been using HiRes only for a few shots so far. Typically this doesn't work for birds, and for posting images to Flickr etc (which is my main purpose) I don't need such high resolution. As for extended dynamic range, I rather do HDR. So I don't have my own in-depth experience. For the very few HiRes images that I have processed, I simply put an extra level of sharpness on top of LR's output using Topaz.