yangxiaobo

We do not use uploaded images for training, nor do we share them with any third parties. We also regularly delete uploaded images, so you can rest assured your data is safe.

When I first started using Google Bard (now renamed Gemini), I hoped to be able to upload my local pdf files for conversation, but I did not find the entrance to upload files and could only upload images. This doesn't meet my needs.

So I started looking for a solution, if I could upload my pdf files in Bard and be able to have a conversation.

After some searching, I found someone saying that you can use an online pdf link, enter it into the input box, and then ask a question. I tried it using someone else's link and it seems to work, which is great. So I uploaded the file to my website, then sent the link to bard, and started trying to have a conversation based on the file content.

It looked good at first, but later I discovered that this function was not really implemented.

Just an illusion. If you use an online PDF file in the prompt word, Bard or Gemini will only answer based on the name of your online PDF document, and then based on its knowledge base or online search for relevant content, not based on the actual content of the document. For example, the name of my PDF document is 'How to use PDF for conversation in bard', but the content of the document is 'How to make a pizza in 7 steps'. When I ask him to summarize the content of the document or answer based on the content of the document, The content of his answer is always 'how to use pdf for conversation in bard'.

So I started a new attempt and finally found the correct way to use it: Extensions based on Google Workspace of Bard or Gemini to talk to PDF files.

Therefore, through this project, I want to introduce to you how to use pdf files for conversations in Bard or Gemini. Share my experience with everyone. Don't be misled by wrong usage methods again.

Hope you enjoy it.

On February 16, openAI launched the Sora model, which seems to have revolutionized previous video generation technology. Whether it's the duration of the generated video (it can generate videos up to 1 minute long, while previous video generation software was still struggling with 4s, 16s timelines), or the consistency of the video (it has solved the problem of inconsistent content and flickering in previous generated videos), there has been a qualitative leap. Especially in terms of understanding the physical laws of the real world, it feels like a technology that can directly generate videos from text, rather than stitching videos together in the form of generating keyframes of pictures like previous models. Everyone is very curious and enthusiastic about everything about Sora, but since it is not yet open for use, we can only see some of the effects of videos generated by Sora. However, the videos are scattered everywhere, making it inconvenient for unified learning and viewing.

So I developed this website.

This model utilizes facial ID embeddings from a face recognition model, enabling it to more accurately capture and reproduce the facial features of specific individuals.

By combining text descriptions, it can generate highly personalized images consistent with the original facial features.

This means that by uploading just a few of your own photos, you can generate images of yourself in various scenarios, essentially cloning your face.

Usage Steps:

1.Upload one or multiple photos. 2.Enter prompt text, such as “A photo of a woman wearing a spacesuit.” 3.Click to Submit

It can also generate stylized images:

1.Upload one or multiple photos. 2.Enter prompt text, such as “a watercolor painting of a woman” or "A sketch of a woman.", or any other style you desire. 3.Switch the Generation type to Stylized. 4.Click to Submit

Additionally, this model can be used in comfyui and sd (Stable Diffusion) as well.

Online Experience Address: https://ipadapterfaceid.com

This is really an interesting feature, haha!

A few small tips.

1.If you want the generated face to look more like the original, upload multiple photos at a time.

2.You can also add some negative prompt words, configure the embedding weight of the face, and the structural weight of the face, for more controllable effects. You can try and experience it for yourself.

3.If an error occurs during generation, try a few more times. Once the progress bar appears, it should work.