1.The Herta (starrail)

2.Elysia (honkai 3rd)

3.Liliya&Rozaliya (honkai 3rd)

4.Syuen&JIEN (NIKKE)

5.Anis (NIKKE)

6.Yvone (endfiled)

ex.reverse bunny

This concludes the regular upload.

I will explain the following along with the announcements.

ex. sound test

These are videos where I attempted to apply sound. I tested various settings and tried changing the prompt. Overall, I think the sound is a bit on the louder side. Please make sure to check your surroundings before playing the videos, and even if you are alone, please lower the volume sufficiently before watching.1. 7 seconds 32 frames (same as existing videos) -> OOM (Out of Memory) -> Failure.

2. 3 seconds 32 frames -> Success -> Approx. 13 minutes

3. 7 seconds 24 frames -> OOM or 2 hours -> Unreasonable to generate.

4. 7 seconds 16 frames -> 15 minutes in fast cases, over 30 minutes in long cases -> Compromise.

5. 7 seconds 16 frames + Reference (trained and applied Yvonne character's voice) -> OOM -> Since OOM occurs regardless of video length, giving up until PC upgrade.

Conclusion: currently producing videos with sound applied at 7 seconds and 16 frames per second, aiming for 10 videos per character. Additionally, since character references cannot be applied, there is a high probability that the character's voice will sound different in every video.

test sample sfw

These files are images I used to study how to utilize AI and how to use prompts (commands).Some nsfw files are also included.

test sample v5 test

The AI ​​model I use to create images has been updated.

Therefore, even if you enter the same settings and prompts as before, the output from the new version of the AI ​​will differ.

For this reason, these images were produced during the process of experimenting to either reclaim my previous style or find a new one.

I plan to hold a vote in the future.