Back in 2017 I shot my feature film Honeysuckle Way on a Canon 600D with a 1.33x anamorphic lense (the Panasonic AG-LA7200G, which was made for the DVX-100). I hadn’t used it for almost a decade and so was interested to get it out and have a new go with it.
At 1.33x the lense turned the 4:3 of the DVX-100 from 1.33:1 into 5.32:3 – which is 1.77:1 or 16:9 — which was just coming in at the time as the desired format, and I guess still is, for TV and streaming.
However when you put that lense on a camera shooting 1920:1080 it turns it into 2553.6:1080 or 2.36:1 or in other words, widescreen cinemascope.
So not having used it for quite a while I was keen to get it out and see what I could do with it. I wasn’t even sure if I could find the lense (eventually did) and it screwed in okay just as it used to.
You can see there’s a couple of rings in between – a stepper ring and a polarising filter with the glass taken out. This allows me to rotate the anamorphic lense into the correct position. It’s all on a rod system and I used a block of wood to sit under the lense hood on the anamorphic lense to keep it square.
This lense is from the long gone and much loved QPIX Centre in Brisbane – I bought it as they were closing down I think. It was hired out to a lot of people so it’s had a bit of wear and tear.
When I shot Honeysuckle Way I used MagicLantern – I think the main feature I was after was zebra stripes – and it worked quite well. I kept it on my camera for years without problems, but then right at the start of COVID I was working at a school as a lab technician and as the kids were all locked down at home I was filming science experiments with the staff for the kids to watch (without the anamorphic lense I might add) – and the camera locked up. That was the first and only problem I ever had with MagicLantern – and fortunately the camera wasn’t bricked and I was able to restart with the Canon operating system – but I’ve hardly touched the camera since then – and so have pretty much forgotten how to use the settings on it.
So let’s look at the results – still images first.
The first efforts were underexposed
And mind you this is before I desquished it. So I had the stills set to 5184 x 3456 to get it to the correct aspect ratio I would have to multiply the horizontal dimensions by 1.33 so this gives 6894.72:3456. Let’s round it up for simplicity 6895:3456 Or 1.99:1 – or basically 2:1. Which is this:
Now that’s pretty ghastly because of the underexposure – let’s try and fix that a little. (Note to self: get it right in camera first – fix it in post is a myth!)
Now that’s still a bit ghastly but at least you can see more. Now because I’m a cinema obsessive I don’t want it in 2:1 – let’s go for something more cinema-like – say 2.36:1 like it would give me on a video frame. Remember this is because the dimensions of a video frame are different to that of a still image frame. Also I don’t think there’s any strict definition of widescreen – I think it’s usually 2.2:1 to 2.4:1 so 2.36:1 sits nicely in there.
So to get this frame to that I need to crop down to reduce the vertical dimensions – so I need to crop my 6895 x 3456 still image to 6895 x 2922 to give me 2.36:1 as my aspect ratio. And on this badly exposed picture, that looks like this:
That feels a bit more widescreen – cinemascope, even with the bad exposure.
So then I went for a drive, setting the camera on auto, until I refamiliarise myself with the settings, and I got the following images – once desquished and cropped etc. You may want to click on them (or right click on them) to see them full size.
You’ll note on the very left and right of frame the image gets a little blurry – I’m not sure what I think of this – does it add to the cinematic feel?? I’m not sure if it’s due to all the lenses/glass – or if it’s because this anamorphic lense has got some kind of damage, after its hard life.
I also had this happen on a few occasions (I won’t desquish and crop the image)
This is vinjetting because I was on full wide – and it’s seeing the lense hood. Using the 18-55 lenses I can really only go about 25mm wide at the most. Which is no big deal. I remember I discovered this early on before shooting Honeysuckle Way.
But still, I like these results – even with the blur – and it feels right when it sits in at 2.36:1 I like that slightly different otherworldly look.
So now let’s look at the situation with video.
The same principles apply in video – but with a few considerations. The basic video frame size, on this camera, is 1920 x 1080 – which will need to be de-squished into 2554 x 1080.
That will give the correct aspect ratio for the pixels and the image will be 2.36:1. However you’re achieving that by stretching the 1920 out to 2554 (remember the anamorphic lense squishes that extra info in). So is that a problem? Yes and no. You’ve got an extra 634 pixels that you’re spreading information across, which is going to create some artefacts and have some effect on the image quality. You’re not making up new information – you’re just spreading the information you have across more pixels than it was originally captured in – these limits will show.
In my opinion you’re better going the opposite way and reducing the vertical resolution – so the image will be 1920 x 814 – but that’s a weird shape for the video file, even though it maintains the correct aspect ratio of 2.36:1. So what you do is you sit that in the middle of a 1920 x 1080 video format — or in other words letter box it (derr!).
This has two advantages – a TV will readily accept this file (I tried a 2554×1080 mp4 on my TV and it wouldn’t play it, although VLC and other media players will on a computer). But a 1920×1080 file is standard whether the image in there is letter boxed or not. And the other advantage is the apparent resolution of your image will be quite good – because you have over sampled – you’ve taken a 1080 high image and scaled it down to 814. This is better than stretching out the horizontal resolution – or creating a fake widescreen effect by shooting 1920×1080 and putting black bars over the top and bottom (which means the effective amount of pixels used to capture the image is greatly reduced).
So to illustrate this let’s use some images from Honeysuckle Way.
This is a still image from one street scene – this is taken directly from the camera original MOV file. This is 1920 x 1080 and anamorphically compressed by 1.33x on the horizontal axis. Note you’ll notice there’s a small vertical black line at the left and right hand end of the image. I’m not sure why that’s there – perhaps it’s like an old fashion overscan area from the CRT days. It keeps the image in a picture safe area. Not sure but the camera places it there and it is definitely part of the original 1920×1080 MOV file that the camera creates. So it can be ignored and just considered part of the image for this exercise.
Next I do a bit of colour grading – not much is needed here because the exposure was pretty good to begin with. Using lumetri colour in Premiere I’ve crushed the blacks a little and given the mid range a tiny boost. This almost makes it looks like reversal film and I like that rich contrasty look. Look at the fence and the grass on the left for the biggest difference.
Now if I was to stretch it out to 2554 x 1080 this is what it would look like. You may want to right click on it and open it in a new tab to get the full effect/size.
However this is what it looks like in the very compliant 1920 x 1080 letter boxed frame.
And here are the three major stages in video. I’ve got to admit those black lines bug me – but that’s probably where the edge of the TV would go, or the bevelling on the side of your phone – hence the image doesn’t go the whole way.
And that’s it. Here’s Honeysuckle Way – the whole film was shot in this manner. I’m currently going back and remastering all my films – and working on Honeysuckle Way at the moment – I think most of changes will be in the audio and a little bit to the colour grading.
And here’s the camera during the shoot (with lense cap on)
