I haven’t uploaded for a while, so I thought I’d show you some of what I’ve been up to.
These are a bunch of clips I’ve merged together, created purely to test things out and find the limitations of my system and resources. There’s been a lot of failed deleted animations I haven’t uploaded too.
You can watch it here on Blogger or watch and download it on Google Drive
Click the settings cog and change the quality to 1440P to view in full quality, and make sure its full screen, it really makes a difference!
My findings.
Resolution size and aspect ratio.
By rendering the non-moving parts of the image with a 16:9 camera aspect ratio. Then changing the cameras aspect ratio to only capture the moving part of the image, in this case an aspect ratio of 5:9 to encompass her moving head. I’ve managed to reduce the pixel count by 2,534,400 per 2K image. Saving a massive amount of rendering time. All I have to then do is paste the one 16:9 image behind the 5:9 image series in photoshop. Unfortunately, it means that the animated parts have to be in the center of the screen. Its worth noting that the clips aren’t merged in the order I created them in. In fact, the kick was the very first rendered image series.
Reducing the video footage from its current 2K (2560 X 1440p) size to HD (1920 X 1080p) would cut the rendering time in half. Creating in HD Ready 720p (1280 X 720p) resolution, which I might do from now on. Would take a quarter of the rendering time as 720P is 4 times smaller than 2K.
Frames per second.
Initially I was rendering at 60 frames per second (FPS). Meaning for every second of footage I had to render 60 images. The kick and where she says “That’s my silence spell” is 60 FPS. The rest of the dialogue is 15 or 20 FPS. The moving camera part is 25FPS, as anything less made it look choppy. I discovered 60 FPS is only necessary for slow motion footage.
After stitching the footage together, I also discovered that you can’t have the video play at different Frame rates. So, it was merged together at 30 Frames per second. Which unfortunately lowered the smoothness where the camera moves.
Using these two methods together I’d be able to reduce rendering time substantially.
I also tried rendering the complete background first, then rendering the figures separately. But in doing so the lighting and shadows are noticeably off. So, I decided not to go down this route. Although it would save a huge amount of time.
Lip syncing
I’m using a Daz3d addon called Anilip 2. It’s supposed to automatically sync the characters lips to your dialogue. Most of the time when it generates the text from your audio it gets confused. Making you have to change the words generated and adjust the speed etc. Using shorter audio clips or smaller portions of text helps it get it right more of the time. I know its not perfect, but I like the results.
Audio
I’ve tried a few different websites as well as MorphVox by ScreamingBee which is standalone software. But nothing seems as good as ElvenLabs. Others have managed to get better results than I can. I’m using their text to speech mode rather than speech to speech as it gives me more consistent voice results. It’s probably what I’m going to stick to for the time being. I’ve considered using voice actors from Fiver. But the prices would soon rack up. I might still go down this route if I find a project that I feel passionate enough about. And if I can find a NSFW voice person who’s affordable.
Sound effects are hard to come by. Most websites don’t allow you to use their sounds for this type of content (Stating you can’t use it for anything that could be considered violent or sexual by another person) even when paid for, something they bury deep within their terms and conditions. But I found a website called FreeSound.Org that seems to allow this type of usage. And best of all its free.
I’ve also invested in a sound effect package; their terms and conditions state you can’t use their sounds to create anything immoral or obscene (Like most content providers). Which personally I think is vague and subjective. I’m hoping my content doesn’t breach these terms.
The Future:
With Daz3d you have to buy an additional interactive licence to use their assets in computer games. A licence that costs around $50 per item, whether it be an earring or a house. Making it practically unaffordable for that usage. And even though you can buy that licence, it doesn’t mean the item is rigged for in game usage, meaning you have to do a lot of things to make it work. However, if you render the asset in video footage and include that footage in a game. You don’t have to pay for the additional interactive licence, you just have to pay for the asset itself. This is because the asset isn’t being manipulated live, and so the game doesn’t contain the asset meshes. In the future, I’d like to make a story driven computer game using Unity and C#, with animated cutscenes hopefully to this quality or better. Where in game choices will make certain scenes load and others not.
For now:
I’ll continue trying to get better, I am still a newb after all. Hopefully I’ll find more time reducing techniques. I’ll try to create some short story animations. I’m still trying to find that format that allows me to create content on a more regular basis. I’m still part way through creating a couple of comics that keep getting pushed back as I find new things like this to dive into. I feel like I’m making rapid progress with finding my way around 3d software, posing, lighting, animating, post production work etc. Becoming a faster, better more competent writer though is a much slower process, it still takes up the majority of my time with these projects. But bear with me. I should get there eventually.
