This is a ZIT series LoRA. I will train a series of photography-related slider-loras to achieve better control over the images. If you happen to need controllable image generation, you can like, save, and follow me. That way, when there's an update next time, the system will automatically notify you.
| Preface
Old players all know that AI pursues visual balance, and it's actually quite difficult to achieve ultimate visual performance. Under normal circumstances, this is still okay, but over time, it's inevitable to feel bored. I've been trying to break through this constraint. This LoRA breaks the inherent lighting balance of AI, allowing you to control the exposure of the image through weights, thus firmly grasping light in your own hands.
If you want to create some highly atmospheric scenes, it will be very suitable, whether it's intense movie scenes...
……or fresh lifestyle inspiration, can all be easily accomplished with just one click!
| Why is it?
Some people may think that adjusting the contrast in Photoshop is enough. Why do we still need a LORA? Let's compare and see.
This is the screaming image on the cover. I made a comparison in Photoshop using weight = 0 (equivalent to the original image) and weight = -3.
Comparison after reducing the color levels of the original image
Comparison after lowering the original curve
Reduced the contrast after underexposing the original image
The effects are clearly visible at a glance. As for me personally, I would choose the effect on the right. The original image will inevitably experience some degree of pixel loss after being adjusted in Photoshop, and there will be some unnaturalness in terms of lighting and color. Of course, if you're a Photoshop expert, then none of these are issues.
Check out my Patreon for more usage tips
| Differences between BASE Edition and PLUS Edition
There are two versions: the BASE version and the PLUS version. The PLUS version is based on the BASE version with adjusted datasets and optimized some training parameters. The following is a comparative test of different weights.
As shown in the figure, the weights are between ±5, and both have achieved the expected results well. With the same weight, the PLUS version is overall darker (or brighter) than the BASE version.
Humans are relatively better at detecting detailed differences, with a weight of -5. Both versions work well. The BASE version has slightly higher image contrast, which is more pleasing to the eye, while the PLUS version has softer lighting. There is no inherent superiority or inferiority between them; it all comes down to personal preference. However, I prefer the PLUS version. When I used to make commercials, the texture of the PLUS version is more like the original film stock, which leaves more room for color grading in post-production.
Under extremely bright light conditions, there will be some differences to a certain extent. The weight is +5, and the prompt is "Portrait of a European woman in the jungle". The BASE version has a certain probability of generating portraits of Asian people, while the PLUS version can better restore the content of the prompt.
| 🙈🙉🙊OK, that's all
If you like the PLUS version → limited-time purchase is now available → but this won't last long. After the purchase period ends, it will only be shared with funders.
Or support my Patreon community, and I will share more results of exclusive research.
Thank you for reading this far. English is not my native language. The above content was translated by artificial intelligence. Please forgive me if there are any inaccuracies.
Finally, have fun! AI never sleeps! Peace!