That is an interesting question. You have to remember that once an image is opened in Photoshop it is no longer raw image data. Photoshop cannot modify raw files. So the smart object no longer contains that raw image data. So, technically, there probably is some loss of quality. But I think it would be difficult to measure.
I'm not a real computer guru and don't know how to measure all this technical stuff. But I just tried opening a raw file as a smart object, doing some changes in Photoshop, saving the smart object, and then deleting the original raw image. I then reopened the smart object, and I cannot see any quality loss. For me, at least, this would not be an issue.
> once an image is opened in Photoshop it is no longer raw image data.
My understanding is that a copy of the original raw file is saved in the smart object. Consequently, when you double-click the smart object thumbnail and return to ACR, you are accessing the raw file that is saved in the smart object.
Are you saying that this is not correct?
That is what I was wondering about, and why I did the experiment that I explained. I really don't know the answer. I know that when you double click the smart object you can go back and modify the data using ACR. The actual editing that is done using Photoshop is not done on the raw image data. That is the whole purpose for the smart object. Again, Photoshop cannot directly edit them raw image data, and cannot save a raw image.
My thinking is that in this situation the smart object that is reopened in Photoshop as a tiff image is what gets opened in ACR. If you have ever experimented with Photoshop/ACR you know that if from Bridge you intentionally open a tiff image in ACR, anytime you subsequently open that image from Bridge they will go to ACR first. And I think that is what is happening in this instance.
Now, hopefully someone who really knows what they're talking about can help us out.
> Now, hopefully someone who really knows what they're talking about can help us out.
Don't claim to be an expert, but I do understand this process.
When the ACR file is opened in Photoshop as a smart object, a copy of the actual raw file is transferred to PS, and PS also maintains its own xmp data for that file. The resulting saved .psd file will include the raw data and the xmp data, any subsequent edits to the raw file WITHIN PS will modify only the xmp in PS, not the original xmp sidecar file.
Most adjustments cannot be performed directly on the image, require adjustment layers, and some operations require that the image be fully rendered.
This is my last comment, I promise. I just looked at the file sizes, and the file size of the smart object is enormous, compared to the original raw file. That tells me that the raw image data is embedded within the smart object. This certainly would not be done as a space saving part of the workflow.
But there would be some quality loss in the image worked on in Photoshop because that image isn't raw anymore. If you can see the quality loss so be it. I can't.
The data saved within PS is still the original raw data, there is no quality loss.
Once the smart object is contained within PS there are two ways to modify the image without affecting the original raw quality:
1. Clicking directly on the smart object layer brings up ACR, and any change made within ACR is non-destructive, and is contained within the embedded xmp data.
2. Adjustment layers can be added on top within PS, curves, layers, etc., which still doesn't modify that original raw data, just its overall appearance within PS.
Pixel based operations require that the raw file be rendered into a normal image. I believe this can be done by creating a new layer on top of the "stack", leaving the raw data intact at the bottom. Pixel based operations can then be peformed upon this rendered layer without modifying the original raw data, however if one goes back into ACR with the object the changes will not be reflected in the rendered layer.
Basically anything useful to be done with the image after editing, i.e. printing or saving out as a jpeg, requires the raw file be rendered into a normal image.
There's no real magic here, just a mechanism for retaining the raw data within the .psd file, basically equivalent to using a copy of the original raw file with a separate xmp.