I wonder if this can be extended to work with general prompts telling the LLM how to behave, such as a DAN mode

Why do you have a fake "See what our users say" section? For example, "Emma Johnson" is https://unsplash.com/photos/grayscale-photography-of-woman-z..., and "Jackson Smith" is https://unsplash.com/photos/man-sitting-near-gray-steel-roll.... The names are generic and the job titles lead to Wikipedia pages

Thanks for your feedback. I've replaced it by a features section

They mostly compare it to the base SDXL. There is only one comparison with a community finetune on page 12 (RealStock v2, I suppose the one from Fooocus) and they don't even link to the model. I would like to see more comparisons with community models based on SDXL since they usually generate better images

No privacy policy, no contact info?

This seems great, but I have 2 questions:

1. Since this is for "early myopia", does it mean it is useless for adults? If not, which groups of adults will find it useful?

2. What do you think about this and is this even related?

> The overall findings are equivocal with under‐correction causing a faster rate of myopia progression. There is no strong evidence of benefits from un‐correction, monovision or over‐correction. Hence, current clinical advice advocates for the full‐correction of myopia

https://www.tandfonline.com/doi/full/10.1111/cxo.12978

Edit to clarify, full correction with glasses should mean always seeing the sharpest picture possible

Thank you, 1. As long as the myopia is progressing there is an opportunity to slow that and avoid complications of high myopia. While screen altering may in principle go beyond the effectiveness of optical devices for myopia control as we can provide arbitrary input to the retinal neural network, realistically myopia reversal for adult humans is not considered achieveable. Intervention should ideally be started as soon as possible. In fact there is the possibility of delaying the onset of myopia or fully preventing it for some future generations, but obviously we are a long way off from there in terms of clinical studies.

2. I think undercorrection is somewhat related. I have not yet read the study you linked, I usually refer to this review for undercorrection. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9213207/ I think it would be possible to achieve some results that way as some of the studies from the review suggest, but in practice it is difficult to precisely control the amount of myopic defocus on the retina that you get that way. Most of the time you get no myopic defocus at all due to accomodation and with strong undercorrection you risk losing the visual cues to the sign of defocus, which results in deprivation myopia. I think this is the reason that MiSight contact lenses are known to be effective, while naive undercorrection has mixed results at best.

May I suggest an anecdote on why some studies have reported undercorrection progressing myopia: as my far sight deteriorated, without correction, I had more incentive to do "near work" like look at screens or look at books (because they were not blurry), than spend time outdoors (because they are a little blurry).