About how to avoid some problems with FluidSimulation 2.0.
The ancient Moon rockets used three-stage rockets to reach space (as seen on then black-and-white TV). A similar three stage approach to avi rendering will simplify things a lot when you're dealing with FluidSimulation 2.0. If your computer is not top-notch, then might a similar three stage approach to even loading the model 流体シミュレーション1024.pmx help you.
1. Let your model move a lot before the release of liquid, but decide that your model's lower body bone must be completely still during the frames while the liquid is released.
2. Add all other models except 流体シミュレーション1024.pmx to your MMD project, and work on it until you know approximately where the liquid should be released. At that point, save your project, i.e. as splashy_video1.pmm.
3. Make a copy of your project, i.e. splashy_video2.pmm. Delete the stage and all models not involved in the liquids action, to make this project ligther on your computer. And turn off RayCast and other resource-heavy effects. Load the model 流体シミュレーション1024.pmx and move its bone Null_01 well above Y=0. Test that you can start the flow. Then go to the frame where the flow should start. Fine-tune the position and angle of Null_01. Register all changed facials and bone settings at the flow start frame. Go back one frame and register zero flow for the facial Null_13. Do the same at frame 0. But first paste the Null_01 bone settings and facials at frame 0. Pretend that you just have made a 'dance motion', and save it to a file like it was a dance motion you intended to share, i.e. WaterSourceBone1_splashy_dancemotion.vmd. And also save this project, i.e as splashy_video2.pmm.
4. Close splashy_video2.pmm, open splashy_video1.pmm and complete the non-splashy parts of your project. When completed save splashy_video1.pmm.
5. Make a copy of splashy_video1.pmm, i.e. splashy_video3.pmm. Open splashy_video3.pmm. Load the model 流体シミュレーション1024.pmx. Load the saved 'dance motion' vmd file for 流体シミュレーション1024.pmx. This is where your computer almost comes to a standstill, right? (Even if MMD on your current computer is unable to display any liquids in the MMD work window, MMD will still be able to render your project to an avi file, thus letting you see the liquids being recorded in the render window).
6. Your desire is now to render the entire video as 60 fps 1080x1920 starting from frame 0, right? If so, render as 30 fps 720x1280 starting from a the first splashy frame instead. Carefully look at the rendered frames passing by, and abort the render when you've seen enough. Go back and improve splashy_video1.pmm and splashy_video2.pmm alternately until your test render of splashy_video3.pmm looks satifactory. (Or if your computer is good enough, then just keep working in splashy_video3.pmm and forget that the two previous versions ever existed?)
7. All done! Let's render as 60 fps 1080x1920 from frame 0? No don't. 60 fps renders will have disappointingly low amounts of liquid. Please render as 30 fps instead. Then let some other video software duplicate each frame for 60 fps. The other software will obviously not reduce the amount of liquid on the duplicated frames.
What is my "other software"? Here is how I'll do it:
0. Always examine the soundtrack at EVERY step in the video production with SPEK, to avoid unnoticed losses of audio quality: http://spek.cc/
1. In MMD load the model.
2. Load the LipSync motion for the model.
3. Load the wav music file to MMD.
4. Listen and watch to check if the vocals are in sync with the LipSync motion. If needed, either adjust the motions' start frame in MMD or the initial duration of the wav soundtrack in Audacity.
5. Load the dance motion for the model. Improve the project bit by bit.
6. Save and close MMD. Rename the wav file with "NON_" in front of the filename, so MMD won't find it. Open MMD and ignore the error message "Cannot find the WAV file". Continue to improve the now mute project.
7. In MMD render to a mute 30fps avi file.
8. In AviUtl open a non-liquids mute 60fps intro avi clip.
9. In AviUtl append the liquids mute 30fps main avi clip.
10. In AviUtl save as avi. In AviUtl will the first clip decide the resolution and bitrate of the entire project. So AviUtl will duplicate the necessary frames needed for 60fps.
11. In VirtualDub open the appended 60fps avi file. Select Video/Direct stream copy.
12. In VirtualDub select Audio/Audio from other file, and then Audio/Direct stream copy.
13. In VirtualDub save as avi.
14. In HandBrake 1.3.3 (or old 0.9.5) open the new dubbed avi file.
15. Select Audio / Audio Codec / AAC (avcodec) / 512 Bitrate; (Or in old 0.9.5 select AC3 (ffmpeg) / 640 Bitrate).
16. If you use 0.9.5, select Video / Target Size (MB) / 298. (Iwara won't accept 299 MB sometimes).
17. If you use 0.9.5, select 2-Pass Encoding and Turbo first Pass
18. Click start to save as mp4 for Iwara or webm/mkv for YouTube. (Webm with Ibopus Opus sounds much better than mp4 with AAC at YouTube. And Iwara only allows mp4, with much better sound quality than YouTube.)
19. If your video is mute because you skipped VirtualDub, then dub for Iwara with Avidemux and select Audio / AAC (FDK) / configure / not lower kbit/s than your original soundtrack (use SPEK). Or just select Copy if your soundtrack already is AAC or mp3 or AC3.
20. Upload to Iwara
21. Skip the VirtualDub part above for YouTube. Instead dub an Ibopus Opus or Vorbis Ogg soundtrack onto your mute webm/mkv file with MKVToolNix.
22. Check the sound quality after the dubbing with SPEK. Certain dubbing methods will reduce the cut-off frequency, i.e. toss away the sound at 18-19 or 19-20 kHz. So test dubbing with more than one method.
The above "other software" method works with 1080x1920 video resolution. And also with 1440x2560, provided that you edit the settings of AviUtl to accept 1440x2560. (Except that VirtualDub doesn't seem to like 1440x2560 video of AviUtl origin over a certain file size, so you'll have to skip VirtualDub and dub the final mute mp4 with some other software, i.e. Avidemux). If you settle for 1080x1920 resolution, then will your selection of softwares increase a lot, because many free softwares are free only up to 1080x1920.
(MMD gives 32-bit colour and AviUtl reduces it to 24-bit colour, which might be sad for RayCast users? If so, test looking for an alternative software maybe?)
Links for audio dubbing with Avidemux:
https://hydrogenaud.io/index.php/topic,100909.0.html (says: use AAC (FDK))