AI Mirror: AI Photo & Video

So many of these types of apps out there—it’s always a gamble to determine what the quality of image conversion will be. And I must say I’ve been happiest with AI Mirror—most especially it’s DIY Lab! I’ve fed my pics and my bf’s into the app and they bear a strong enough resemblance in both cases—although maybe a little more “idealized” and whitewashed than I’d prefer—but most definitely us. (In my case, the Lab enlongated my short neck and hasn’t quite captured my upturned nose.) A slider in the Lab would be helpful in terms of having the option to create a more faithful interpretation vs one where more liberties are allowed. That being said, I use the DIY Lab to create characters based on photos and, in those cases, I don’t mind the slight variations as I don’t want exact replicas of the original pic. What I do love is that the character generated is consistent from rendering to rendering, and continues to look like the same person each time (with an allowance for the occasional AI hiccup). What I would like to see more of: diversity of skin color, hair texture and body type. There’s already been some improvement, so I know this is something they’re working on, but it’s not coming fast enough for those of us who want to feature darker skin, tight curls, natural hairstlyes, and thicker bodies. For women especially, I’ve noticed a tendency for the Lab to lighten skin tone and relax curls, and it would behoove the developers to understand that whitewashing is not a feature but rather a bug, and one best remedied sooner rather than later.

Hello, I really have enjoyed this app very much and the way I can save images, but the new interface has changed a lot of that and it’s difficult to use and I have to go through extra steps to save it to the location I want. I miss the “original vs new comparison” design from the old interface too, and liked how it had a tiny thumbnail of the original image in the corner. Also, when I go to save an image to a specific location, before I can choose options it automatically saves the image in the default form first before showing me the alternate options for where and how I save it. This is very annoying and is just a waste of space on my phone since the default save setting converts it to a document image file, which won’t show the thumbnail when I post it to Discord and will instead be in the form of an embedded file to download. There were some other features I enjoyed such as the option to watch an ad to make an HD rendering of the art a certain limited times a day as a free alternative to otherwise having to pay for unlimited HD renderings with the subscription. Please bring this feature back! It was really nice (just to clarify, I haven’t subscribed or done the 3-day free trial before) but now I haven’t been able to watch ads to make an HD rendering anymore since the interface change. This new interface is overall more confusing and I dislike the layout. Please do something, whether that means going back to the old interface with the features I had enjoyed, or making lots of changes to the new interface and bringing back some of those free (meaning without a subscription) features that the old interface had. Please fix this issue, this is very inconvenient and it makes it difficult to enjoyably use the app. I really do like this app and would like to continue using it. Thank you for your time.

First thing first: when it works, the results can be pretty awesome. The problem comes with the “when” part. If you’re taking selfies and running them through, most of your results should be fine. It’s when you try to do something else that things fall apart. AI has always had a hard time with hands, but seems it also has issues with cats, cell phones, hats, and anything between your legs (clothed or not). If your picture contains any of that, you are just as likely to get some Eldritch Horror as you are to an AI art piece. It was also fairly obviously trained with mostly female examples because it likes to default to putting lady parts on anything it can’t figure out. I’d also say that the “unlimited” images is pretty much a straight up lie. I paid $30 and still get “wait 5 mins and try again” messages. I have been using it heavily trying to see what does and doesn’t not work, but if you say “pay x amount to get unlimited images” then one might expect to actually process unlimited images without interference. So it’s a fun app and, with the right input, can produce some wonderful and engaging images. But step off the path and you’re going to hate the results. Some of the filters NEVER seem to give decent results, but having a Favorite tabs helps you keep the ones you like in one place. For a beginner AI app, it’s serviceable. Side note: app vastly over estimates how many men have tummy tattoos versus body hair….also, if you want to process video, that’s $2 a video, no matter how it turns out. Avoid the videos like a plague.