My Experiences With Project Odyssey

In The Beginning
When I heard of Project Odyssey for the first time in early 2024, I read all about it and felt reluctant to submit anything.

You see, I only discovered A.I. at the start of 2024, and being very new to it felt that my submission would be a waste of time. I felt I didn`t know enough to submit something worthy of the judges scrutiny.

I had seen what AI was capable of by seeing posts on Instagram and CivitAI, and I really felt that I was going to be a small fish in a big pond. And, for days I felt a sense of unease about deciding against entering, at times almost sad that I had given up before I even started.

A little bugger sitting on my shoulder taunted me for days "Just do it, what have you got to lose!?"

So, I decided to enter.

I had some experience making images in LeonardoAI, albeit using very simple prompts. Most times I relied on their prompt enhancer to make better images, and I thought "My images are the greatest thing since sliced cheese!! Everyone is going to watch my little video and they are going to LOVE IT!!"

I jumped in... armed with my 1060Ti and 8Gb of system RAM. I began learning about AnimateDiff, what the heck a LoRA was, what on earth a ComfyUI was, and why everyone was using it? I need a python on my computer now?! Whut?

So many questions.

And for every question, a thousand YouTube videos existed to answer them.

For almost a month I spent all my free time outside of work creating "Tired and Wired", my very first full AI music video.

I used ComfyUI workflows that others had put together, consulted multiple YouTube channels to learn from them, even though most times I was like "What on Earth is this person talking about?!!" and I had to re-watch the same video a number of times before something stuck in my brain.

And man, were there days when I just wanted to call it quits and go back to making cat pictures on Leonardo.

My PC would take 8 to 12 hours to render a single AnimateDiff video. I didn't have enough money to buy subscriptions to all the latest AI sites popping up. I was watching all this cool stuff being made by others and posted to their socials, and I was just a poor guy with a crappy machine, and no skills.

However, I persisted.

I kept going until I had uploaded the finished video. I counted down the days to when the winner would be announced, secretly hoping that my little video would be the Music Video Category winner for Project Odyssey Season 1.

Well that didn`t happen. Not. Even. Close.

As the other entries started piling in, my jaw spent most of the time on the floor as I watched the God-like levels of submissions that came drifting in across all the categories. "Holy hell!!" I thought to myself, "These people really know what`s going on, there is no space for me here!"

I didn't come even close to so much as a mention, but something had bitten me.

Infected me.

I was addicted to this AI thing. In all it`s forms.

I was going to get better. I was going to take this somewhere. I would one day be victorious in this contest.

Enter Project Odyssey Season 2

"This is it!! I am better than I was the first time around!! I have been practicing the dark arts late at night, the tools are better, the models have become stronger, my entire PC rig has been upgraded to a 4080 16GB, 64Gb system RAM, hell yeah let`s gooo!!"

And so the cycle began again. Day after day I spent creating what was in my head, prompting on CivitAI using locally available LoRA`s, downloading models to run inside of ComfyUI, as I was now able to use it more effectively!!

I was rendering in Hailou, lip-syncing in Kling, writing songs in Suno, I thought I was a CapCut pro and editing like I was some Hollywood big-wig.

I uploaded my beautiful creations to the contest, and went and had a Merry Xmas and New Years holiday with friends and family. I felt confident that I had done enough to at least garner an honorable mention, if not an outright win!!

Well that didn`t happen. Not. Even. Close. Again.

And so began the process of "What went wrong?"

"Why were my videos not chosen?"

"Were they so bad?!?"

"Should I just give this whole thing up?"

You see, the thing is, whilst I had leveled up from the first contest, so had everyone else. And so had every AI model. And as everything else leveled up, so did the judging criteria.

That bugger on my shoulder piped up again, "See what others have done. Look at what they did. Look at what you didn't do."

So, I looked. And I compared.

And here it is, my list of things I feel need improvement, in order to take my creations to the next level.

Get A Better Editing Tool

This is not a bash on CapCut at all, I absolutely love the product.

It has served me well in the past, and helped me take the ideas in my head and piece them together for all to see.

However, I discovered that it is actually quite a rudimentary tool for professional editing, and that if I wanted the quality I was seeing submitted by others, I would need Photoshop and After Effects, Topaz, and maybe a few other tools. Pricing is sure an issue here, what else can I use?

I started looking around at the alternatives and have since learned I need to become a DaVinci Resolve expert. I see that it is an extremely powerful tool, what with all the abilities it has, especially with color correction and audio manipulation.

Color Correction? Audio Manipulation? What the hell is even that?

These are terms I was not familiar with, but in the last week or two I have been going through a bunch of DaVinci tutorials, and my lack of knowledge in these two areas most likely played a massive part in my works not making it.

There is something I have come across called Invisible VFX, that play an extremely vital part of what makes one video great, and another award winning.

Invisible VFX are basically things you put into your creation that are not visible to the viewer, sort of. If you watch a video and a car blows up in the most spectacular fireball you have ever seen, you go "Whoa that was some awesome VFX!!"

What you don`t physically see, sort of, is the color correction to make the scene more realistic in line with the story being told, the music and sound effects in the background, the character dialog, the voices used, the volume of things, all this falls into Invisible VFX. You can`t normally see it, but it affects how you interpret the scenes you are watching. Well, this is my understanding of this right now. It`s all very confusing at this point.

Audio is another big contributing factor. In most of my time spent on AI tools that output audio, I have noticed a lot of compression - another new word I have learned and need to investigate further. Too much compression on an audio track often distracts the viewer from enjoying a scene, especially when dialog and the background music feels and sounds a bit thin.

I began watching YouTube videos on DaVinci Resolve, and I hear you have everything you need to produce a quality product with the free version, and using the paid version for more advanced editing. This sounds like a better approach moving forward when it comes to editing.

DaVinci Resolve is not automagically going to make my creations better at the push of a button. There is a lot of work to be done here.

AI Is Just Words

Yup, in the early days everyone I knew was going on about "AI is so easy, you just type some random words in the box and voila, you are now an AI Prompt Engineer!!"

This is an entire subset of skills I need to work on, especially my prompting with the video models. I made the most amazing images but then the animations fell apart due to a skill issue in not being able to address the video generators correctly. Sometimes, the model really did not obey my instructions though, even though they were perfect for the animation I required.

In order to achieve greatness, I need to spend more time refining the skills I already have in this area. Prompting is most definitely at the top of my list of areas to improve. It really is the point at which everything starts.

Upscaling

As the saying goes, garbage in, garbage out.

My video quality leaves much to be desired in my view when compared to other videos I literally salivate at. Even after upscaling my images in ComfyUI, then upscaling the videos I got out of Kling and Hailou I made from those images, I feel the quality is not up to standard.

Little glitches here and there can make or break a scene, and blurry images can disrupt things totally. Improving upscaling is one low-effort yet high-impact checkbox for me to tick in my workflows going forward.

Judgement Day

I have created something. I have put my brainchild out for the world to see. It is on the Submission page on CivitAI, my Instagram profile, and my YouTube channel. I sent it out to my friends and family.

In the context of this contest, my creation was uploaded to be judged. Not only by my peers, friends, and followers, but primarily by a panel of judges most qualified for the job.

I accept their decision without prejudice because I do not know any better.

Note to self: Take it on the chin. Apply introspection. Find the faults, and do better next time.

And the Winner is......

Me!

Well, not exactly a winner as in placing in the contest, but the following happened on my socials as a direct result of entering Project Odyssey :

CivitAI - Over 500 new followers gained, and new friendships forged. I`ve also received offers from others to work together in future, to assist me in taking my creations to the next level.

Instagram - I finally pushed through the 1000 follower mark, something I had been struggling to achieve for over 3 months.

YouTube - My music video "Merry Mayhem" has been viewed over 29,000 times since I uploaded it, and has gained me 298 new followers at the time of typing this, accounting for over 850 watch hours. Phenomenal.

These gains are massive for me, and I will gladly take them as a win.

To all the creators whose works have been accepted into the finals, congratulations!!

Your works have no doubt inspired others to do better, as they have with me. The announcement for the winners is due to take place in a few hours, and I sincerely hope and wish for the AI Gods to be in your favour.

Thank you all for the inspiration you bring to this contest.

Wrapping Up

Project Odyssey is a way for me to partake in something that pushes my creative boundaries. To throw myself into the lion`s den, forcing me to get better.

I have seen a lot of incredible submissions in Project Odyssey, each and every one I have watched has in some way inspired me to keep on pushing forward with this.

The Community over on Project Odyssey`s Discord is something else as well.

The one thing I have seen that sticks out the most for me, are the creators chatting about banding together and helping each other out, to in future tackle this beast known as Project Odyssey.

Mark my words, there is going to be some mindblowing work coming out of these collaborations.

The next big movie studio is waiting to be born, and I`m all in.

See you in the next one!