I got this product because I wanted a quick way to be able to take existing product shots and apply my brand aesthetic to in order to give all the product images on my website the same look and feel to make it all cohesive. What really drew my attention and pushed me to the “purchase” side of the scale was the fact that you can train the AI on a brand identity and it would then be able to intuitively apply that knowledge to any image. Unfortunately, once I got everything set up, I discovered that in order to use that particular function you need way more credits than this plan gives, and unused credits don’t roll over at the end of the month so I would have to shell out more money in order to be able to use that function. I live in Canada so the deal already cost a little less than double what the advertised price for the product was — so spending even more was out of the question.

I tried to figure out a work around, and with the help of the built in Phygital AI assistant (which is awesome and you should use it whenever you can), I managed to set up a work flow that should hopefully achieve a similar result — unfortunately, it did not. I spent almost 800 credits in the span of an hour trying to get a usable result, but not a single image that it generated for me was actually usable — forget “usable” from my brand perspective…I’m talking usable period.

Each image generated kept coming back with these very unnatural, odd colours and tones overlayed on my product along with backgrounds I didn’t want (I had asked to have a simple, just black, background with no textures, but kept getting wrinkled shirts, closeups of a man’s chest in a grey shirt, a grey brick wall, etc).

I tried changing the prompt, I added negative prompt, I specifically said do not alter the product itself and just change background, I swapped out the reference image, I swapped out the product image for a completely different one, I set the AI creativity level to 0 — each result came back the same. I even deleted the workflow and started fresh….same result. I consulted the AI for a better prompt and to help with tweaking the settings…same result. I tried externally consulting Claude and ChatGPT separately to help with settings and to figure out what was wrong … same result.

After an hour I gave up in frustration.

To be fair, I have not yet tried to generate an image from thin air, as I mostly just tried the functionality of editing an existing image. Nor have I tried the video nodes or other specialty nodes that are offered, so I don’t know if it’s any better at generating from scratch compared to editing an existing image.

The workflow of the product is also something that takes a minute to figure out — it’s not that it’s difficult, it’s just different than any other generative AI that I’ve used before so I wasn’t used to it, and it’s not intuitive. You built actual workflows by dragging and dropping nodes and images onto your workspace and connect them to one another to create the generative flow; then you add in your prompt and alter your settings to control the output. Honestly, if it weren’t for the AI assistant I probably wouldn’t have figured it out as quickly as I did — so, my advice, go directly to the AI assistant right from the start and ask it to walk you through the steps to do whatever you’re looking to do. It will save a lot of menu opening and clicking on your part. Also, Make sure you have the best possible settings and prompts from the beginning because each time you tweak them it’s gonna cost. The initial generation is around 80 credits, depending on the node and tool you use, but the subsequent regeneration’s were about 365+ credits each time — this is actually what contributed to me going through so many credits in such a small timeframe (although in this instance, it wouldn’t have made a difference if the settings were perfect from the start because the AI just didn’t want to do what I was asking — who knows, maybe it was grouchy or had been having a bad day, or something).

Overall though, I do think this can be a great product — IF I’m ever able to get it to generate a usable image. I’m going to go back and try out the rest of the functionality and hopefully I’ll have better luck with those nodes. I am disappointed, though, that I can’t actually do the things I initially bought this product for — but hopefully I’ll be able to find a workaround or get the current workflow working at some point. Until then I’m stuck burning through my monthly credits experimenting to find out how to get better results.