Disco Diffusion (DD) is a Google Colab Notebook which leverages an AI Image generating technique called CLIP-Guided Diffusion to allow you to create compelling and beautiful images from just text inputs. Created by Somnai, augmented by Gandamu, and building on the work of RiversHaveWings, nshepperd, and many others.
“Fix your gaze on the black dot on the left side of this image. But wait! Finish reading this paragraph first. As you gaze at the left dot, try to answer this question: In what direction is the object on the right moving? Is it drifting diagonally, or is it moving up and down?”
By stimulating specific cells in the retina, the participants claim to have witnessed a blue-green colour that scientists have called “olo”, but some experts have said the existence of a new colour is “open to argument”.
The findings, published in the journal Science Advances on Friday, have been described by the study’s co-author, Prof Ren Ng from the University of California, as “remarkable”.
(A) System inputs. (i) Retina map of 103 cone cells preclassified by spectral type (7). (ii) Target visual percept (here, a video of a child, see movie S1 at 1:04). (iii) Infrared cellular-scale imaging of the retina with 60-frames-per-second rolling shutter. Fixational eye movement is visible over the three frames shown.
(B) System outputs. (iv) Real-time per-cone target activation levels to reproduce the target percept, computed by: extracting eye motion from the input video relative to the retina map; identifying the spectral type of every cone in the field of view; computing the per-cone activation the target percept would have produced. (v) Intensities of visible-wavelength 488-nm laser microdoses at each cone required to achieve its target activation level.
(C) Infrared imaging and visible-wavelength stimulation are physically accomplished in a raster scan across the retinal region using AOSLO. By modulating the visible-wavelength beam’s intensity, the laser microdoses shown in (v) are delivered. Drawing adapted with permission [Harmening and Sincich (54)].
(D) Examples of target percepts with corresponding cone activations and laser microdoses, ranging from colored squares to complex imagery. Teal-striped regions represent the color “olo” of stimulating only M cones.
In the retina, photoreceptors, bipolar cells, and horizontal cells work together to process visual information before it reaches the brain. Here’s how each cell type contributes to vision:
DiT360 is a framework for high-quality panoramic image generation, leveraging both perspective and panoramic data in a hybrid training scheme. It adopts a two-level strategy—image-level cross-domain guidance and token-level hybrid supervision—to enhance perceptual realism and geometric fidelity.
The trigger phrase is “equirectangular 360 degree panorama”. I would avoid saying “spherical projection” since that tends to result in non-equirectangular spherical images.
Image resolution should always be a 2:1 aspect ratio. 1024 x 512 or 1408 x 704 work quite well and were used in the training data. 2048 x 1024 also works.
I suggest using a weight of 0.5 – 1.5. If you are having issues with the image generating too flat instead of having the necessary spherical distortion, try increasing the weight above 1, though this could negatively impact small details of the image. For Flux guidance, I recommend a value of about 2.5 for realistic scenes.
DISCLAIMER – Links and images on this website may be protected by the respective owners’ copyright. All data submitted by users through this site shall be treated as freely available to share.
