Encoding or Transcoding – what do they mean and which is best?
It's a question that's been asked many, many time before – What's the best video encoding format that provides the best results for the web? This is something I've spent countless hours of research on and continue to develop over time. As the internet changes so indeed do the steps required to ensure decent encoding results online.
For ages the battle of file size vs quality has always caused issues when streaming high definition video online. For many, watching such a high bit rate video online is still not possible due to the extremely poor bandwidth so many people still experience from their so called 2mb, 4mb or even 8mb and 16mb broadband connections. I'm one of those people who live in sticks and my broadband connection is rubbish. (Personal Gripe Coming Up). It annoys me why I have pay the same monthly tarrif as someone who lives in town whos getting a cracking 8mb connection, whilst I'm just a measly 4 miles from the local phone exchange, yet I rarely crack 1mb download speed. My phone call quality is the same as it is in town, so why shouldn't my broadband be? Should be based on a pay per bandwidth-quality type thingy, I'd gladly pay top dollar for a 16Mbit connection any day of the week, but I do seriously grudge paying for something I'm not getting! (Ok, rant over – sorry about that).
Anyway, back to encoding video. I wont bore you to death with results and figures of every single render I've done over the last 6 months, but I will try to share with you what I found to be a winning workflow combination and I hope you can take something from it and save yourself a headache and a shed-load of time! During my tests I reckon I've rendered close to 300 videos, in 7 various formats and transcoded them testing around 10 or so different pieces of software, so some serious painful testing time Here goes…..
I use a Panasonic HVX200 camera for most of my shoots and usually capture in 720pn with a frame rate of 24fps. This combination works really well for me as it allows for sensible recording times across my P2 cards and reasonable HD size image for DVD and or web distribution. Dropping the Panasonic MXF files into Vegas via the DVFilm codec is a nice, quick and easy workflow. I can dump 32 GB of HD video into my NLE and be ready to edit inside around 8 to 10 minutes. I couldn't go back to DV tape now – a 1hour tape meant 1 hour to get it into your computer. Ouch!
I tried so many different output renders from my NLE (Sony Vegas) that it was sending me round the twist trying to come up with the ideal solution for each format. I found my best result was to render out the best looking, most colourful, high contrast version I could possibly achieve and then use a video converter to transcode the video file into the formats I required depending on the videos final destination.
I rendered out some test videos from Vegas as MP4, MPEG 1, MPEG 2, WMV, AVI, M2t and MOV. They all produced vastly different results depending on several video content factors. Movement and transitions in my guitar tutorial videos were minimal, but I still experienced some choppy pixelation across large blocks of a single colour when rendering to the likes of MPEG 1, MPEG 2 and MOV. I wanted to test as many formats as possible to try and find a happy medium between file size and quality. I also filmed in front of a green screen for my teaching videos, so I had edge artifacts, lighting and colour correction to deal with throughout the video.
Here's a screen grab of a before and after the chroma key. There is quite a bit of colour correction and contrast adjustment applied to enable rich, deep blacks and a smooth keying out the screen.
Below is the Original Raw Image
Below is the finished image with Chroma Keying, Colour Correction and increased Contrast with a smidge of Sharpening.
I'm really pleased with the results I'm seeing on my screen and I hope it looks okay for you guys too? You can clearly see an increased contrast between the two images as the black t-shirt pops nicely in the rendered image, but what is interesting is that the skin tone has barely changed between the two images, yet it looks bucket-loads richer in the bottom image. I reduced a little of the red highlights from the face and that was about it . This is because when the blacks are truly black and not a washed-out grey/black like in the raw image, the colours really pop ten-fold and a perceived saturation in all other colours start to appear.
I could continue for another 10 or 20 paragraphs or so, but this post is already long enough. I'll post Part 2 very soon and discuss the final format I mastered to, what software I used and how I converted the mastered videos into top quality and sensibly sized formats for web distribution.
Bye for now!