Your use of light and shadow is excellent. In fact, I found that DALLE is not inferior to MJ, SD. It just has some problems causing its development to be lower than it should be. There are also negative changes taking place. For example, these pictures
This picture on the left is a real person. The right is created from DALLE using the prompt created from GPT-4, interpreted from the first image and slightly edited. I created it when DALLE3 was first added to GPT-4. I don’t have any understanding and building skills yet. You will see that if you add a little bit of edited text it will be very realistic.
People are excited about SORA that OpenAI has launched, it’s been talked about and waited… But not for me
The release of SORA for me is DALLE adding more capabilities and adding parameters to the calculations that only increase the variance of the results. SORA use Interpretation of text from DALLE^30. SORA is not perfect.ect.
People lack of understanding of AI behavior and DALLE usage techniques. They still lack the ability to develop techniques knowledge and understanding. Most people like to use prompts like those used with MJ, SD. You can take those prompts and write them by cutting out some words. But it still has the same effect. Or maybe it will get better …In developing techniques, it was found that DALLE was underdeveloped in many abilities. Among those capabilities, there are features that other AI text to image do not have or are different, such as creating letters. Can be further developed. In addition, DALLE also has awareness of space, dimensions, and distance. From testing it was found that If developed further, it would be possible to use layout-based interior decorations, but this is still a long way off. Because it is not a wide range of applications and can be done directly, and there are also more difficult factors because we cannot communicate directly. The only way to do this is to communicate with a prompt to create an image, letting the AI know what it is creating. It’s like creating the e that DALLE knew about but refused to create. Even now the shape has been almost completely created. But definitions such as lowercase cannot be used in construction. But it is the first step that is considered a prototype for the next step. a b d f g h i j k l m n q r t u y
This section is a walkthrough of what I’ve observed and summarizes the implementations that add to the lack of development. The problem isn’t with creating the prompt, it’s with RLHF. We have to make the AI understand what we’re creating by explaining it and stating that this is what we want. And it takes many people to help.
Clear and appropriate DALLE development section In order of importance to be confirmed, they are as follows:
- To practice and enhance skills in using DALL·E for creating images with complex descriptions involving direction, coordinates, dimensions, and distances, the following ideas can be employed:
1.2. Experiment with Perspectives and Dimensions:
- Practice describing scenes from various viewpoints, such as “a top-down view of a building” or “a 3D perspective of a room.”
- Create descriptions that transition between perspectives, like “shifting from a frontal view to a side view.”
1.2 Utilize Directions and Positions:
- Practice using specific descriptions to indicate directions or positions, e.g., “a cat sitting on the left side of the room, facing the window.”
- Generate images with multiple objects placed in distinct positions and describe their spatial relationships.
1.3 Explore Distance and Proportion:
- Practice describing distances or proportions, such as “a large tree standing 10 meters away from the house.”
- Create images that show the difference in size between objects, like “a giant cat sitting next to a skyscraper.”
1.4 Experiment with Complex Detailed Descriptions:
- Use descriptions that include direction, coordinates, dimensions, and distances in a single explanation.
- Create complex scenes like a cityscape viewed from above or an interior room with many organized objects.
1.5 Pay Attention to Light and Shadow:
- Practice describing the effects of light and shadows on objects and scenes, like “light from the window casting a long shadow of the cat on the floor.”
- Create images with varying light and shadow to observe their impact on perception of dimension and distance.
-
Experimenting with Blending Art Styles:
Experimentation with blending different art styles involves combining features or techniques from various art movements to create new, unique artworks. For instance, merging the emotional and atmospheric use of color from Impressionism with the dream-like, reality-bending elements of Surrealism can produce images that evoke emotions while presenting themes or objects in unexpected ways. -
Exploring the Use of Color and Texture:
Creative use of color and texture can dramatically alter the mood or meaning of an image. Experiments can include using unconventional color schemes, such as employing blue to denote warmth instead of the traditional red or orange, or introducing unexpected textures, like giving water a paper-like surface. These experiments add new dimensions and depth to artistic communication. -
Applying Technical Details:
Paying attention to technical details such as lighting, shadows, perspective, and dimension can enhance the quality and comprehension of an image. Skills in these areas allow for the creation of images with greater accuracy and depth, such as using lighting and shadow techniques to add a sense of dimension and distance, or adjusting perspective to emphasize the importance of objects in the scene. -
Creative Experimentation:
Encouraging free and creative experimentation is crucial for developing skills in using DALL·E. This could include creating images that blend reality with fantasy, using symbolism and metaphor to present ideas or emotions in new forms, or challenging audience expectations by presenting objects or scenes in unexpected contexts. Such experiments can open new avenues in artistic communication and increase the ability to produce meaningful and influential artwork.