AlwaysOnAuto
Well-known
Clear as mud, count me in!
Proposed RFF AI Policy
- Thread starter splitimageview
- Start date
- Latest activity Latest activity:
- Replies 29
- Views 724
sojournerphoto
Veteran
Very supportive of this and the suggestion to disallow the use of members’ photographs and writing to train Gen AI models.
Keep it human.
Ray Kilby
Established
I assume that those who use the most recent photoshop and the Ai inbuilt tools are not the target here. I have cloned to remove blemishes or artefacts I dont want for years; and zoned my areas of light to balance an image. (Like we did in the good old darkroom days.) I've also interpolated and dare I say it, altered focus on rare occasions, added grain, layer images to drop in different exposures. Is this acceptable? If so, if someone drops an image in to say Gemini, or any other version of Ai and prompts it to do any one or more of these things, is that not acceptable? I don't disagree that we have to be mindful of what is posted, and creating a policy is a good idea, However, do you think it might need further nuancing? Maybe a 'guideline' might be better than a policy. I'm not be sophistic here by the way.. How about a panel, if this is to be seriously investigated to be made up of administrators, regular members and maybe an expert on post production even someone who has some skill in ethicacy? And they could provide a very comprehensive guideline. What do you think Splitimage? Is this too much?
Coldkennels
Barnack-toting Brit.
I'm not @splitimageview, nor am I a mod, but I do think you're overthinking this.
The simplest way to put it - being as we're photographers - is that if you can't do it in a darkroom, it's probably veering away from photography and moving into "digital imaging", and therefore isn't appropriate.
You could layer and composite images in a darkroom. You could increase or reduce grain. You could airbrush or paint out items or imperfections.
You can't, however, alter focus or create whole things which weren't there. That's not photography any more. That's not "drawing with light". That's getting a computer to generate things for you.
splitimageview
Administrator
Photography, after all, is the art, application, and practice of creating images by recording light, either electronically by means of an image sensor, or chemically by means of a light-sensitive material such as photographic film. - New Oxford American Dictionary
If the final image was created/modified by a prompt, it's simply not photography.
generative AI is what the proposed policy is directed towards:
• text-to-image, invented content, subjects that may never have existed, was not photographed, adding generated elements to a genuinely photographed subject
Image processing is simply executing math. No training or 'intelligence' despite what software publishers decide to label it.
• that is: curves, color balance, dodge/burn, sharpen, contrast, white balance, saturation, noise reduction (of the strictly mathematical type - Gaussian blur, bilateral filtering, etc)
computational photography may use ML but it starts with a real (not an invented) photo/subject.
• masking, subject selection, face detection, synthetic bokeh, enhanced noise reduction (that is, Topaz Denoise, latest Lightroom, etc.)
Hope this helps.
If the final image was created/modified by a prompt, it's simply not photography.
generative AI is what the proposed policy is directed towards:
• text-to-image, invented content, subjects that may never have existed, was not photographed, adding generated elements to a genuinely photographed subject
Image processing is simply executing math. No training or 'intelligence' despite what software publishers decide to label it.
• that is: curves, color balance, dodge/burn, sharpen, contrast, white balance, saturation, noise reduction (of the strictly mathematical type - Gaussian blur, bilateral filtering, etc)
computational photography may use ML but it starts with a real (not an invented) photo/subject.
• masking, subject selection, face detection, synthetic bokeh, enhanced noise reduction (that is, Topaz Denoise, latest Lightroom, etc.)
Hope this helps.
splitimageview
Administrator
I think the important distinction isn’t whether AI is used, but how it’s used. AI-assisted editing tools (denoise, masking, depth estimation, etc.) still begin with a photograph of a real scene. Generative AI creates subjects and content that were never photographed.
Personally I have zero interest in viewing someone else's prompt creations. This is a constant annoyance on fb and IG; scenes that were never photographed, events (videos) that are complete fabrications, in an effort to garner clicks and coin. I would hate to see that annoyance proliferate on RFF.
Personally I have zero interest in viewing someone else's prompt creations. This is a constant annoyance on fb and IG; scenes that were never photographed, events (videos) that are complete fabrications, in an effort to garner clicks and coin. I would hate to see that annoyance proliferate on RFF.
agentlossing
Well-known
I think this is as clear as it can be, and all of the comments raising what ifs seem to be ignoring where this line is being drawn.
Godfrey
somewhat colored
I presume that last bullet point includes "... does not alter a member's original wording or meaning ...". 😉
Sounds good. I come to this and any forum to have discussions with and see the work of photographers, not trade 'AI' generated fuff.
G
ChrisPlatt
Thread Killer
Richard G
Veteran
This is great. I’ve always considered we would be super-detectors of AI generated text. Conversely, we have now found an authenticity justification for some of our second rate photographs.