Hidden

November 3, 2023

by GymDreams

Sometimes not seeing something can be just as powerful as being able to see it.

Hidden. Sometimes not seeing something can be just as powerful as being able to see it. Stable Diffusion with Virile Reality.

Leaving things to the imagination can often make a higher impact. When you don’t spell everything out, it lets your mind run wild, giving you a rush of curiosity and creative thinking.

I created these images a while ago to showcase the influence of suggestion in visual content. While I understand that most people lean towards seeing everything depicted, I personally find it more captivating to leave certain aspects to the imagination. Maybe it’s just my own preference, but I can’t help but wonder if there are others who feel the same way.

In the past, I had a blog catering to a gay audience, where I would share pictures of men. As an experiment, I tried two different approaches: one set of photos that revealed everything and another set that hinted at things. Surprisingly, whenever I asked my followers in a poll which they preferred, they always chose seeing everything. However, I noticed that the images where things were merely suggested actually received a much higher number of likes compared to the ones showing everything. Based on these statistics, it seems like most men might not be aware of their own preferences.

To achieve this using txt2img technology, the key is to avoid explicitly describing certain elements. For instance, I discovered that instead of specifically requesting a full shot of a man, focusing solely on his sneakers allows you to position the camera in a way that directs the face out of the frame. While it’s still possible to remove or crop out the face in post-processing, it’s generally more effective to accomplish this during the rendering process — the quality of AI renders is highly sensitive to the resolution of the image, so trying to maximize what you can do directly is highly preferred.

Stable Diffusion txt2img, with Virile Reality 1.0. Steps: 25, Sampler: DPM++ SDE Karras. ADetailer: face_yolov8n. Hires upscale: 2, Hires steps: 10, Hires upscaler: R-ESRGAN 4x+, Automatic1111 Version: v1.3.2. Post-processing: Topaz Gigapixel, Adobe Lightroom, Adobe Photoshop.

Rendered: 2023-06-14
Published: 2023-11-03