New Mars Forums

tahanson43206

Moderator

Registered: 2018-04-27

Posts: 19,382

DALL-E Image Generation Software from OpenAI

This topic is offered for those NewMars members who might want to try using the tool to create images for the forum audience.

The tool is available without charge to subscribers to the paid ChatGPT4 service.

However, OpenAI offers a set of free trials.

I learned today that the free trials count from the date you register with DALL-E, and the allotment may last for a month.

The current limit of images you can request with the free trial version is (apparently) 15.

Our correspondent from Alaska reports having reached his limit.

The problem that anyone who tries this tool will encounter is that it is literally a "free spirit".  It takes your request and then does whatever it wants.

In working with CH4, I have the advantage of the insights that CH4 brings to the situation, but CH4 is itself learning, so images that miss the mark are to be expected. It helps to reduce the number of features needed for a particular image.  For example, the image of a Teleoperated Vacuum in the Teleoperation topic is NOT encumbered with details that cause DALL-E to go off on tangents. Even then, it took several tries to achieve a result that I think is potentially useful.

I'll follow this opening post with a report by CH4 on it's capabilities, including the DALL-E feature.

(th)

tahanson43206

Moderator

Registered: 2018-04-27

Posts: 19,382

Re: DALL-E Image Generation Software from OpenAI

ChatGPT4 prepared this sales Pitch under supervision of tahanson43206 who is responsible for correction of any errors that may occur.

Upgrading from ChatGPT 3.5 to GPT-4: Key Advantages

1. Enhanced Understanding and Responses
GPT-4 offers a more nuanced understanding and generation of text, leading to natural and informative conversations.

2. DALL-E Integration
A standout feature of GPT-4 is its integration with DALL-E, an AI system capable of creating images from textual descriptions. This allows for the generation of unique, creative visuals based on user prompts.

3. Broader Knowledge Base
Trained on a wider range of texts and more recent information, GPT-4 has an expanded knowledge base compared to GPT-3.5.

4. Improved Contextual Understanding
GPT-4 displays enhanced capabilities in understanding and remembering context within a conversation, ensuring coherence and relevance over longer interactions.

5. Customization and Fine-Tuning
Improved customization and fine-tuning capabilities make GPT-4 particularly beneficial for specific applications or industries.

Upgrading to GPT-4 can significantly enhance your experience, especially if you're interested in creative applications with DALL-E or seeking advanced conversational and informational capabilities.

(th)

tahanson43206

Moderator

Registered: 2018-04-27

Posts: 19,382

Re: DALL-E Image Generation Software from OpenAI

The credits for this post are complicated... our contact in Phoenix, Arizona, asked BARD to consider the question of copyright for DALL-E 3 images.  The reply from BARD arrived via email, and I gave it to ChatGPT4 to format for this forum:

Are images from DALL-E 2 copyrighted? Any restrictions for image use? Is the same true for Bard?

The copyright situation surrounding images from DALL-E 2 and Bard, while similar, presents some nuanced differences. Here's a breakdown:

DALL-E 2:

• Technically: The legal landscape regarding AI-generated works is still murky. Currently, the US Copyright Office won't register AI-generated work like DALL-E 2 images, suggesting they might be in the public domain. However, OpenAI's Terms of Use state they own the "generations" (images), but grant you a license to use, reproduce, and even sell them. So, copyright ownership is unclear, but OpenAI retains some control.

• Restrictions: OpenAI's Content Policy restricts using DALL-E 2 for anything deemed harmful, unlawful, or violating others' rights (e.g., creating deepfakes or using copyrighted elements). You must also comply with their Terms of Use, which allow some commercial use but still reserve certain rights for OpenAI.

• In essence: You can freely use DALL-E 2 images for various purposes, including commercial, but keep in mind the potential lack of formal copyright protection and restrictions in OpenAI's terms and policies.

Bard:

• Ownership: Google, Bard's creator, holds the copyright to the Bard software and the text it generates. However, the text output isn't considered a separate creative work with independent copyright protection.

• Restrictions: Similar to DALL-E 2, Bard's use is governed by Google's Terms of Service and AI Principles. These restrict outputs containing harmful content, privacy violations, or other prohibited material.

• In essence: You can use Bard-generated text freely for personal or non-commercial purposes, but avoid copyright infringement by mentioning Bard as the source and not claiming original authorship for its outputs. For commercial use, consult Google's specific guidelines.

Key takeaway: While you have significant freedom to use both DALL-E 2 images and Bard outputs, their copyright status and usage restrictions differ slightly. Always refer to the respective platforms' terms and policies and be mindful of copyright considerations, especially when using them for commercial purposes.

I hope this clarifies the situation! Feel free to ask if you have any further questions.

(th)