Latest posts

As we get closer to “Computer Enhance” there’s a big problem

Artificial Intelligence driven image uprez has been improving dramatically over the last few years. There are already commercial products like Topaz’ Video Enhance AI that seem like they will inevitably lead to the “computer enhance” featured in so many science fictions shows, where it is used to pull some tiny detail out of an image, or a license plate, etc.

Ironically, that’s probably one thing it cannot be used for.

Google have showcased a new approach whose development had been sidestepped in 2015 to allow other approaches to uprezzing to be explored. It seems like the original idea developed to fruition can create amazing results. From a Cined.com article:

How does that work? Well, the team took a high-resolution image and added noise until only pure noise remained. They then trained a neural network to reverse the process in order to recover the initial image. Finally, this trained neural network is used on images at 64×64 pixels to create superscaled versions. Jonathan and Chitwan then used this process in a stack to further improve the process. By stacking a 64×64 ? 256×256 model with a 256×256 ? 1024×1024 model, they were able to achieve impressive upscaling results.

https://www.cined.com/new-google-ai-image-upscaling-makes-science-fiction-a-reality/

Making super resolution out of noise seems like alchemy to me. It will most certainly have many uses, particularly incorporating legacy material into more modern productions shot at higher resolutions.

As for the “computer enhance” problem. The “detail” is being invented, based on what the AI thinks it should look like. For realistic images this is excellent. For enhanced detail you could rely on for detective or legal purposes, almost certainly not. The detail is, quite literally, made up!

Thoughts on Frame.io and Adobe

Now I’ve gotten over the surprise of the sale price, it really doesn’t matter what I think. What it comes down to this:

To Adobe, Frame.io is worth $1.275 billion. Ultimately that’s the bottom line. Adobe invest in anything they think will add value to the company and pay what makes sense for them.

I’ll admit, the ascribed value is close to double what I would have expected for a Frame.io exit, and that was a bit of a shock. What I wasn’t taking into account is that something created and working, is vastly more valuable than the best idea ever. Frame.io took an idea and executed, creating an entirely new workflow in Camera2Cloud.

I’m sure Camera2Cloud is where Adobe see a lot of the value, and rightly so. Elegant review and approval is a neat achievement, adding C2C has truly created something new and elegant. As technology develops what is currently a rarefied workflow will become more and more accessible. I can see why Adobe would want to be leading there.

No matter how talented your engineering team is, there’s no guarantee that an Adobe team could have replicated what Frame.io have already completed. Going it yourself, has inherent time-to-market disadvantages, and an ongoing problem playing “catch up” with the pioneer.

I can’t pretend I understand how all those variables are ascribed a value to get to the final deal, but I don’t have to! The deal was done and both parties are happy.

So, unqualified congratulations to Emery Wells and the Frame.io team.

EditMentor sets a new standard in educating future editors

Up front I should state that I have been a fan of EditMentor’s founder Misha Tenenbaum’s work since he first founded EditStock.com. I was excited when he first hinted at what he intended with EditMentor, although I did not envy the challenge.

You see, I have a little insight. For a large part of my career I was an educational designer, and came up with the concept and content for our series of “Intelligent Assistants” – ultimately for Final Cut Pro (classic), Boris Red et al, Cleaner and Media 100. Our goal was simple: let people learn how to use a NLE or similar tool, while working in that app.

That took significant work, but by comparison with what Misha and CTO Gor Vardanyan have done it was nothing. Teaching functions in an NLE is easy. Teacher people why they should use those fi\unctions is not easy.

EditMentor is based around a functional NLE, that runs entirely in a browser (specifically Chrome). A NLE interface that is smart enough to report clip, marker and trims to the back-end for scoring! Add in a student management system, and a complete set of tools for building lessons, EditMentor is an amazing technological achievement.

Except, I think that’s not EditMentor’s main achievement. Working through the challenges never feels like learning. The challenges level up slowly, and feel “easy.” Well, they are easy when you read the Challenge and don’t jump ahead thinking you already know what to do! Who would do that? Oops.

EditMentor doesn’t care. Miss an answer, and there’s the explanation as to what is correct. You can go back and retake a challenge after learning the answer, because it’s all about learning. Still uncertain? Go back and review with all the answers displayed!

When learners have the freedom to fail without consequence, they want to learn. If I needed to go over a Challenge a couple of times to work it out in my head, I’d have all the freedom I need.

From an Educational Design perspective, the lessons are well set out, with each Challenged contextualized, before being presented with some “rules of thumb” to use in approaching the Challenge.

Challenges use the full range of NLE tools: navigating Bins, Editing to the Timeline, a Range of basic editing and trimming tools, and a functioning multitrack timeline. Wherever the student progresses, they have a good foundation to take forward to working in any NLE. The Lessons and Challenges are well designed.

EditMentor’s challenge is to going to be to fill out the curriculum, which they are acutely aware of and working hard on.

Although EditMentor is working with some of the top educational institutes there is a free option for individual students. Combined with the readily available phone cameras and editing tool, there’s no reason anyone with a desire to learn editing skills to not learn.

This “quick” review has taken me far too long. I keep feeling that I should add more to it, but each time, I find myself simply repeating myself, raving about how good EditMentor is. If you want to learn, or your role is helping other people learn how to edit video, then you must check out EditMentor.

Rotoscoping and Infill tedium gets a new Machine Learning Replacement

In the back of my mind I always thought, if I had it to do over, I’d go into Visual Effects. The combination of creativity and technology that goes into VFX fascinates me. Then the reality of the industry that I read about and its financial woes, makes me rather glad my life took a different path.

But what would VFX be like if the tedium – rotoscoping and background removal – were simply a couple of clicks away. The video below compares “traditional” (up to Rotobrush 2 in After Effects) rotoscoping tools, against RunwayML for isolating moving actors, and in filling. Stick around to the end for a preview of ML creating cartoon images based on text descriptions!

Machine Learning drive RunwayML is a web service, also used in this real time – 10 minute – demo of rotoscoping three separate clips and exporting the results.

These incredibly powerful tools are only the beginning of how Machine Learning is transforming almost everything visual: from production tools like RunwayML, to deep fakes like the recently distributed one of Tom Cruise.

Machine Learning is making huge strides in almost everything visual. It’s a pity it’s not advancing as fast at deriving useful keywords!

These tools are, however, the way I see Machine Learning (a.k.a. Artificial Intelligence) integrating with human creatives to devote more time to polishing and finishing the effect, to to process effects shots that would never be done otherwise. That last part is the exciting part for me.

Subscribe to The Corridor Crew for a bunch of interesting VFX videos.

Or for more of the applied techniques ActionVFX for the “Best Stock Footage for Visual FX.”

Reviewing Larry Jordan’s “Techniques of Visual Persuasion”

When Larry Jordan asked me if I knew anyone who might like to review his new book, Techniques of Visual Persuasion I immediately volunteered myself. After all who doesn’t want to “Create powerful images that motivate?” Plus, I always feel I can learn more about marketing and motivating people.

The first chapter of the first section – Persuasion Fundamentals – sparked off ideas as I was reading. It seemed like every new topic inspired and idea I could use to improve my marketing.

There is so much good information in the first chapter alone that I’m scheduling time to reread and apply the lessons.

The balance of the first section takes the principles from chapter one and demonstrates how it applies to create persuasive writing, images, fonts and colors. It’s a nice balance between the theoretical challenge and practical application.

The first section was the most valuable for me, because it was the newest material. The book is a nice balance between the techniques of persuasion and the techniques of creating persuasive writing, images and video. A marketer can rely on the practical production techniques, while the visual creator can learn from the techniques of persuasion.

In the areas of the book where I have some expertise, it was reassuring that none of the basics had changed! Larry carefully, and methodically, draws the connection between the theory and why (and how) it applies when you pull out the lights and cameras.

With such a visual subject, you would expect a well designed book, and New Riders delivered an all color, very handsome work. The illustrations are spot on and make understanding the “point” comfortable.

An unexpected, and welcome, inclusion is a chapter on Persuasive Presentations. Personal Presenting is a skill I continually strive to improve, so I devoured that chapter and will take a few new skills to my next presentation. (Perhaps Larry can write an extra chapter “Persuasive Zoom Presentations!)

The ‘Persuasive Practice’ for that is Pecha Kucha (or “little talk” in Japanese): a 20 slide, 20 seconds per slide presentation. Taking up the challenge I created a presentation of 20 Lessons from Learning to Sing. (I took up singing as a mature adult about six years ago.) It was an excellent challenge for me. Creating a precise 20 seconds from my verbose starting point was good practice.

The presentation has now become a 20 day series my vocal coach will be publishing later this month.

Techniques of Visual Persuasion is full of useful, practical information that both marketer and creative will find compelling, packaged in an entertaining and easy-to-read format. If you want to improve your persuasion techniques, and who doesn’t, then I thoroughly recommend Techniques of Visual Persuasion.

Disclosure: I asked to review Larry’s book and was provided with a review copy from the publisher. I received no other compensation.

Content Creating Entrepreneurs

I do not immediately think about satirist Randy Rainbow as a Content Creator, but with a body of parody music videos that have hundreds of millions of views combined, he is obviously a successful Content Creator.

As an aside, I am totally jealous of his ability as a lyricist, but that’s another story.

In a recent CNN interview Randy Rainbow talked through the process he takes for each new parody music video. It turns out he shoots and edits himself, on Final Cut Pro X. (FCP X is for those who create for themselves, etc.) He makes his income from producing videos, so is he a “professional?”

There are not many performers with millions of followers/views that have complete control over their career. Certainly not in traditional media, but for those who want to create content finding their own niche and audience isn’t as impossible as it was to get to the top in “Hollywood” and be a star that way.

Sam Mestman weighed in on what’s needed to be a modern content creator in a recent article at FCP.co.

The Top Ten earners on YouTube earn between $15-22 million a year as Content Creating Entrepreneurs. Musical artist Troye Sivan started building his audience on YouTube when he was 13 living in Perth, Western Australia (not the center of the entertainment world). He just had his best year ever with a World Tour and new album.

When you bring the audience you end up with a lot more negotiating power when it comes to distribution, merchandising and marketing deals.

Not just content creators but entrepreneurs as well. But are they “professional” editors? Definitely not. Professional Content Producers, definitely.

We have a 10 Year Old App!

While it’s not the first app we released, Sequence Clip Reporter has now been serving the Final Cut Pro, Premiere Pro and Resolve communities for 10 years! That’s an impressive achievement for a small Independent Software Vendor.

Greg and I are fortunate that we are able to make a decent income serving our community with tools that give them more time to do what they really want to do, instead of the boring parts of post. We’ve seen other ISVs move on to employment where they can get a more stable return for their skills. Being an ISV is a hard business.

Sequence Clip Reporter arose from a suggestion by our friend Les Perkins after I demonstrated some (now defunct) apps Greg had written. In those early days we had no idea what software helpers people needed, so we tended to follow up on most suggestions. Sequence Clip Reporter was an exception as most ideas turned out to fit a fairly narrow workflow.

Sequence Clip Reporter was created for FCP Classic, and we expected it to fade out over time, but with a little tweaking to accommodate Adobe’s custom version of FCP Classic XML, it’s now a valuable tool for the Premiere Pro community. It turns out Sequence Clip Reporter is the better of our reporting tools for Resolve as Resolve is a track based NLE.

Of course Producer’s Best Friend is our FCP X reporting tool that fits in where Sequence Clip Reporter does for Premiere Pro and Resolve. While superficially similar – they both produce all types of reports – the structure and the way the two apps work is remarkably different.

Happy Birthday Sequence Clip Reporter.

The Terence and Philip Show Episode 85: Is It Time to Retire?

The discussion in this episode started with a recent article in The Atlantic Your Work Peak Is Earlier Than You Think. Is it time to think of a career as an evolving series of “careers?”

After listening to the show, what you you think?

Has editing digitally reduced our ability to visualize?

When I think back to the days of editing tape to tape you had to both visualize how the series of shots were going to edit, but you also had to keep in mind and plan the current, previous and next edits at once. It was hard to go back and “trim” an edit! It usually entailed remaking every edit from that point!

With digital NLEs we make the changes instantly. We can go back five clips and trim an edit to make it flow better, without needing to reconstruct subsequent edits.

That flexibility has made it so much easier to edit visual stories. I certainly don’t miss those days, but has it reduced my ability to visualize the whole sequence, or doesn’t it matter?

Instantly preview and edit without that creativity killer: preroll alone is a huge step forward!