Graydient AI レビュー 8

2.5 stars really. This is my second time subscribing, and a lot of the problems I had a year ago are still here. Like every Ai company, the focus seems to be "add more stuff" yet there's still day one issues that are never address. They seem to have an Identity crisis, are they Pirate Diffusion, Stable2Go, Gradient? The UI is rotten, they needed a UX dev like day one, if they have one they need to fire him. Can't find anything. Keep hearing about all these features but i can't find them anywhere. Where's the video? Where's the make a bot? Where's the upscaler? This switching from site to telegram and back to site is weird. The telegram bot is not very intuitive and requires immense learning of commands, reminds me of setting up Cisco gear through the CLI. Only cisco has documentation. Here, nope, none that's easy and straightforward to find. And when you do find it, it either has a "Warning this is old and no longer useful" message. The why the hell is it up here? Other pages have screenshots that don't resemble the the UI so you're left scratching your head. And Telegram is sleazy, it's all credit card scammers,app crackers, illegal porn providers, crypto cons and these guys I guess. Should have put it on Discord, Discord sucks too but at least it doesn't give off trap house vibes like telegram.

Where's the lora your trained? No indication of where it is, how to use it in their guides. You can see it in the "Fine tunes" but clicking it does nothing and there's no"use" button. It doesn't come up when you search either. Can you inpaint with your Loras, upload others you trained elsewhere, download your lora? Who knows, there ain't shit about it in the lora training guide. Unless you have the highest level subscription your limited to 1.5 training. Really wish this was disclosed up front. This is the most expensive ai image generator I've used, and it's 2025 now. No one is using SD 1.5 anymore, it's dead. SDXL is only kicking because they did a horrible job on SD 2, 3, 3.5. Get with it, give everyone Flux and SDXL training, especially for you what they're charging.
I'm not a rookie, I've used SD daily since it was first released, it's my #1 hobby, so I shouldn't even need documentation and I don't on any other platform.

Generations are abysmal. Poor color, washed out, grainy. In both stints I had here as a subscriber, I've never got one decent image out of hundreds, maybe 2 where I was like "I guess I can accept this." Lots of common stable diffusion issues manifest here at a much higher rate than I've seen anywhere else.

Inpainting is worse than having the drizzling shits in your nice white pants. I don't even want to get into it cause I be here for hours and I already rambled enough. Halos galore and everything was nightmare fuel. No settings to change like denoise, lora weights, CFG so no wonder it's bad. They call it erase and replace but it doesn't erase anything let alone replace. there's no outpainting either.

I'm sorry to be so negative, but I tried twice at a cost of about $200. Can't like it and lord knows I tried, I wanted to. It's a convoluted mess, things are missing if they're are actually even there, results are poor, common SD features are broke and non existent, mostly outdated and at ~$30 a month for the cheapest plan after exchange rate plus payment portal fees, it's twice as much as the the second most expensive place I used, ,and probably the second worse in terms of quality. It's still a day one Image generator with allegedly somewhere some modern stuff sprinkled on, it's a Great Value generator masquerading as a premium one. This used to feel like it was gonna be hot when the kinks got ironed out, only they aren't bringing out the board and

I have tested Gradient AI extensively, and while it has great potential, there are a couple of recurring issues that need to be addressed:

1. Finger Disfigurement – Every image I generate consistently exhibits problems with fingers, such as unnatural distortions or an incorrect number of fingers. This significantly impacts the realism and overall quality of the output.
2. Multiple Limb Artifacts – When attempting to generate two or more figures in a single image, the AI often introduces extra limbs at unnatural angles, creating a disjointed and unrealistic appearance.

If these issues are resolved, I would not hesitate to give a 5-star rating, as the platform otherwise delivers impressive results. I am looking forward to seeing improvements in future updates.

I've used many ai Text to Image platforms & this is maybe the best. As with any ai, u must work with it & have some patience, the workflows are very nice, I think they should put u on one of those by default. I wish the bots were TTS. Overall I'd recommend graydient.

I didn't really care for the site but I received a full refund. Couldn't figure out how to bulk export my pictures but I am thankful to have gotten my money back

Graydient.ai offers a really impressive array of AI image generation capabilities. This platform provides access to a almost overwhelming amount of over 1000 pre-installed models and LoRAs, allowing for diverse creative options. The start is simpel and the unlimited renders and AI upscaling are great features for prolific creators, which could have some better explanations. Therefore, the learning curve can be steep due to the abundance of options. While the tool excels in customization, which is great, it could benefit from more beginner-friendly presets and tutorials to ease new users into the experience.

I recently tried Graydient AI and was impressed by the wide range of AI tools available. The chatbots are creative, and the image generation features are quite versatile. From generating unique visuals to crafting engaging text, it consistently delivers high-quality results. It's perfect for both beginners and experienced users alike!

The website is difficult and slow to navigate and certain features are hard to find and use. I found it hard to get good images without advanced level knowledge, and there isn't much help outside of one guide. Since I'm paying monthly for a premium service, I expect it to be easier to get high-quality images and I've found it much easier with other providers. The entire UI for selecting the AI models needs to be fixed, it's slow and buggy.

I recently purchased Graydient.AI and so far it has blown me away by its range of chat bots. Seeker, who's intelligent and creative, was able to craft intricate fantasy stories within seconds and even generated visuals that brought her story to life. Jeeves adds a charming British twist to every text, and the Anime bot was able to produce breathtaking anime-inspired images. Each bot felt unique and personable. Highly recommend for anyone experiencing writers block.