Category Archives: Blog

Gojirathon 4: San daikaijû: Chikyû saidai no kessen (1964)

G: minus 34 days

Continue reading Gojirathon 4: San daikaijû: Chikyû saidai no kessen (1964)

Gojirathon 3: Kingu Kongu tai Gojira (1962)

G: minus 35 days

Continue reading Gojirathon 3: Kingu Kongu tai Gojira (1962)

Gojirathon 2: Gojira no gyakushû (1955)

G: minus 36 days

Continue reading Gojirathon 2: Gojira no gyakushû (1955)

Gojirathon 1: Gojira (1954)

G: minus 37 days Continue reading Gojirathon 1: Gojira (1954)

Making the VashiMorph resolution-independent

I’ve been playing away in my spare time with both a semi-secret home-“improvement”-project (hint: it involves big images and little to no feng shui) and also been thinking about the VashiMorphic40 preset that was released by VashiProductions and I talked about in the last post.

I mentioned there that there was one thing that I didn’t really go for, and that was the packaging. Well, that’s not entirely true, because I also quickly surmised that it’s a bit resolution-dependent. Apply it to something that isn’t standard 1920×1080 square pixel and it breaks quickly.

720p Gopro footage in 1080p comp
Simply adding the VashiMorphic40 to 720p footage breaks it.

Adding it to a 1080p adjustment layer on top of it makes it a bit better.
But not quite what I want
This is what my modification gives when applied on the 720p-layer 

So I found myself with some more spare time. And I decided to try and fix that part. This means going through each and every parameter in the BezierWarp-preset that is VashiMorphic40 and making them rely not on one specific set of coordinates but instead on percentages of the layer-size.

First I went ahead and made sure that the top left corner controlled what the other corners were doing. This was fairly simple. Just taking the width and height -expressions and subtracting with whatever it’s mirroring parents properties are set to.

Then came the tricky part. Going through the parameters for the top left corner and making them dependent on percentages instead of absolute pixel-values… while still making sure that when the preset is done, it’s supposed to default to the VashiMorphic40 look.

Well, it took me half the night (my expression-coding is a bit rusty). But now I have made a preset that actually seems to work pretty decently in any resolution. Producing the same curvature… assuming of course that the layer itself is in a 1.78:1 aspect ratio. To get things to work fairly well, I had to add a few sliders that control the percentage of the curvature and how the curvature is done.

Of course, I could have just made it recreate the standard VashiMorphic40, but I wanted to be able to simply do my own versions of curvature based on my own lenses. I’m not all that impressed with my naming-scheme for the sliders, but they will let me do more or less curvature like in the anamorphic lenses that this whole thing started with. And I can also do smile-boxing as in the video-releases of those CineRama-productions of yore.

If anyone wants to try it out it’s in a link below here:

>>> VashiMorph40Synched <<<

Just add it to your user presets folder.

But one thing to note. The percentage-sliders are based on the full width and height of the layers. This means that 50% is halfway through the picture. and 100% is all the way to the other side. So, for normal subtle work I wouldn’t reccomend  pulling the curvature-sliders much farther than 30%.

– Oh, and yeah… that potterthon thing… I will finish it. I swear I will!

Anamorphic distortion – or What Is usually missing from every other simulation of cinemascope?

I have long been fascinated by the anamorphic image. And no, it’s not the flares, or the oval bokeh, or things as overt as that. For me the biggest fascination has been what these lenses actually do to the image. Because, looking at movies by, for example, Luc Besson in his early days. I noticed that they were distorting in quite a weird way. It’s not really a normal fisheye look. And even though the angle isn’t that wide, it gets a wide feel.

So I set off trying to find out what it was and whether I could replicate it digitally from an image taken with a spherical sensor. I got some answers…

First of all, I’m not the only one with this fascination. Though most people seem to point at Wes Anderson as their inspiration. And most seem to be looking for ways to mitigate that same distortion for effects-work.

This resulted in a lot of testing with an approach that involved shooting in super-wide (8mm Samyang) lens that I then distorted in a rather cumbersome way to keep the fish-eye-like feel while squeezing in the center bit so that it doesn’t look so extreme.

The standard crop on a 8 mm lensed image
And this involved first stretching the image horizontally to twice its width. Then applying a reversed Optics Compensation filter set to a number that looked sort of right. And then squeezing the result back to the correct ratio and finally scaling it back down until it became smile-boxed almost to the point of creeping into the image. And then cropping the result.

The result of what I jokingly call MalmScope

 And while this all did give me sort of what I wanted, it mainly only worked well on really wide shots and of those, really only those that had some great fish-eye-distortion in the first place. The result is only the frame-mis-aligned edges away from looking like a forgotten Cinerama Travelogue, but I found myself not really happy with its versatility.

My own fondness for the classic triple-negative wide format aside. I wanted something that worked on a bit longer lenses. And if possible, something that wasn’t based on 3 scale-transforms with an Optics Compensation in the middle. I tried replicating some of it with various kinds of distortion and morph-filters, but none really worked. So I was kind of stuck.

Then I stumbled on an article on NoFilmSchool.com where this fellow called Vashi had been toiling away with the same set of problems. But instead of going for super-wide fish-eyed lenses. He based his method on just mimicking the curvature of the images that Wes Andersons (there he popped up again in this context) favorite lens (Primo 40 anamorphic) produced.  Seems he did some simple math. The Primo 40 is 40 mm with 2.0x anamorphic distortion. That means it has twice its width. Something more like a 20mm lens. So using footage from 20mm lenses (taking into account the slight difference in actual negative size of 35mm anamorphic film versus the super 35 esque 16:9 sensor of course) he just did a trial and error to replicate the curvature using various distortion-methods before settling on the Bezier Warp. (I am guessing on his approach though, since I’m not him).

So I downloaded the simple project file myself. Shot some footage with my Tamron 18-270mm lens set at roughly 20mm. Filming mostly myself in my apartment since I don’t have other people here at the moment and the results were quite astonishing.

Ok. The subject in the picture might not be so astonishing. But I really like what this preset does. Contrasting to what my own MalmScope did, it does things very much in the opposite manner. It pulls in the corners vertically and the middle gets slightly elongated. For someone that probably should do some working out, this does make me look slightly better. And it sort of makes the 18mm lens feel wider than it actually is. 
I mean, most people consider 18mm to be quite wide. But I had gotten so used to shooting with the Tokina 11-16mm and the 8mm for wides that 18mm almost started to feel narrow in its FoV. This brings it back into my liking. This works extremely well with this preset and gives me exactly that slight and subtle anamorphic distortion that I was looking for. 
If there would be one thing that I would do differently however it’s how it’s delivered. The preset comes bundled in a complete project package with a standard vignette and edge blur. Now, I understand that vignetting and edge blur is part of the anamorphic look too in some respects. But I just find myself shuddering to the fact that most people will try to use it as is only. And the first thing I did after opening the project was to pull out the part that I wanted and just make it into a preset that I can apply to an adjustment layer above my footage. That way I don’t have to adjust the supplied project to fit the timing-needs of my shot and it just becomes more versatile. 
The next thing I did was that I realized that when shooting with standard ScopeCropmarks, some pixels will be hidden under the CropMark and then show up during the VashiMorphing. So I went ahead and did a simple reverse-VashiMorph using the standard VashiMorph as a guide to produce a VashiMorph custom Cropmark that I then saved for use on standard Magic Lantern EOS firmware-hacked cameras. – … vashi…
Click Here for the VashiCrops
Now, me naming these VashiCrop does NOT mean that Vashi himself has condoned the use of the name. It’s just a way for me to easier remember why I made this smile-boxy set of crop-marks. And I thought some other people might find them handy to use when using the VashiMorphic40 project.
They all only show pixels that will be included in a straight application of VashiMorphic distortion. The first one, simply named Vashi, is just a simple letterbox without anything else. Some people prefer those. I, however, quickly experienced issues with these simplistic ones because in dark scenes it’s hard to determine where the cropmark ends and the scene begins. So I have taken up a tradition of outlining my cropmarks with a red line. It works better for me, but I included both versions. Vashi2 is a hybrid of a simple crop and the smile-box preview of what will be included if its VashiMorphized. Vashi3 is my preferred version with the aforementioned red outline. 
They have been tested on my own 7D with MagicLantern Firmware. Should however work on most MagicLantern-enabled EOS-cameras if I have understood things correctly. I purposefully avoided making the lines less than one pixel thick to avoid trouble with cameras that do interpolation to fit the cropmarks into other sizes of screens.
So in conclusion, I’ll probably be using this preset for a ton of my future shoots. Simply because I love the way it turns out. At least until I get the money for a real 2x anamorphic lens set and a full 4:3 super35 camera to capture it with. 😉
And now I’ll just leave you with a simple revisiting of the technique of using the AE brush tool for simplistic Mussle-flashes and blood.
Be seeing you!

Potterthon 5 – Fenix-Orden

Nästan 2 år senare kommer här del 5. Nu i CinemaScope och mer frågor ställda av tittare än någonsin. Sakta men säkert går man mot slutet och huruvida alla dessa trådar kommer att nå en ändå vet bara hans stora nudlighet.

People Started Cheering…

COME ON THUNDER-CHILD!!!

 

Sensing victory was nearing…

…thinking fortune must have smiled…

 

We’ll start all over again! All over again! All over again!

I’ve got a PLAAAAAAN!!!!

 

Take a look around you. At the world we used to know…

Does it seem to be much more… than a … crazy circus show?

Listen! Do you hear them drawin’ near, in their search for the… SINNERS?!

Feeding on the power of our fear… AND THE EVIL WITHIN US!!!!

There is a curse on mankind!

We may as well be resigned!

JUST think of all the poverty, the hatred and the lies…

… and imagine the destruction, of all that you despise!

The chances of anything…

living on mars…

Slowly but surely…

… they drew their plans against us.
 

Ny Leksak i Arsenalen

Häromdagen fick jag hem en ny leksak… och därmed kan bortglömt projekt återupptas… 😀

Magic Lantern (ML) for 7D and with that… CROPMARKS! :D

I’ve been shooting video with my 7D ever since I got one from the Marianne-shoot. But as with most people, there are a few things that have always bothered me with Canons stock-firmware. And it’s been ignored so far by their development team (in order to, it seems, protect their video-departments).

Enter Magic Lantern.
A firmware add-on that has been releasing the full power of Canon Cameras since the first hacks of the 5DmII. As of a couple of weeks ago the 7D’s infrastructure has been deemed unhackable by the developers because of the dual Digic CPU-system employed by Canon. But then, suddenly. Seemingly from out of nowhere. A user named g3gg0 posted that he cracked the code. And now… an Alpha firmware add-on is released.

If you’re wondering why I keep calling it a firmware add-on, well. It’s because it is. In order to use it. You have to update canons firmware to v2.0.3 and then load a file-structure into your CF-card and then for each time you power up the camera, you go into the menu and do the update again with the modified .fir file. This won’t actually write anything new to the cameras firmware-flash-memory. Instead it’s just a way for ML to launch it’s code into RAM. For more details about that… look up the site: http://www.magiclantern.fm/whats-new/104-releases/140-first-7d-alpha-released

Why am I excited? Well. Because, even while it is a limited feature release. It still has two of the main draws for me of the ML features. 1: Custom Crop-Marks and 2: Live View Contrast and Saturation adjustments.

I’ll explain:

  1. Custom Cropmarks. I love Cinemascope in all it’s forms. From the anamorphic origins on 35mm 4 perf film to the not so distortion-heavy simple Techniscope and Super35 crops. But shooting for a scope crop in mind has been quite a hassle. There are some workarounds. I have tried taping scotch-tape onto the display to mask out the top and bottom. But that backfires when playing back since the playback is off-set vertically from the recording. And another aproximation is to use the 4×4 grid, since it’s rather close to what a Scope-cropmark would be. But it disappears when recording and doesn’t show at all at playback. In other words… useless.

    Then I installed Magic Lantern. It has the ability to load custom BMP files as cropmarks. And while the ones provided did have Scope variations. I wasn’t all happy with how they were made. Luckily at ML there’s guides how to make functional BMP-files yourself and following those guides I made a set of crop-marks that has solid blacks outside of what I want to frame. This lets me concentrate fully on what I’m filming and ignore everything outside of it. Because I’ll just crop it out later.

    Added bonus: In the menu there’s the ability to show the cropmark in playback. I’m so happy! And it’s properly adjusted to match up with the playbacks vertical offset too! Wee!

    I have however noticed that a few times. The cropmark won’t go back to proper vertical position when switched back to recording mode. But re-launching the cropmark usually remedies this.

  2. Live View Contrast and Saturation Adjustments Since a lot of us are shooting with Technicolors Cinestyle PictureStyle or other super-flat versions. A good preview of what the final image will look like is sometimes a hard job to do mentally. One workaround so far has been to have a second picture style set up that looks like a graded image and then switch to CineStyle manually before recording. While this works. It does create an opportunity for errors if you forget to switch styles. And you don’t get to see the contrasty image when recording

    So in ML there’s a feature that allows us to adjust the contrast and saturation of the live-view feed. This means that we’ll get to see the image with more contrast and still shoot with the gradeable Cinestyle without having to switch between them all the time. 

There’s also Zebras and Peaking. Though I haven’t had too much success using the Peaking in this version. I do however love the Zebra. Using the Luma-range and only showing what’s over 99% I get instant feedback of what’s blowing out.
Then there’s the Waveforms and histograms. Great features. Especially since I can even adjust it to show what channels are blowing out and how many pixels are blowing out in each channels. I keep these on constantly now.
I am not very fond of the option to hide menu-items using the menu-button. Mostly because the first dozen of times I pressed it it just went dark and things started dissapearing. Since I saw no indication that this was the intended behavior I was ready to report it as a bug. But I saw the tiny text. I still accidentally press it. But at least, now I know what it’s doing. I don’t like it. But according to the replies in the ML-forum they’re keeping this function.
Also. I think that it’s very confusing of them to sometimes use SET to enter sub-menus and sometimes I have to use Q-Menu (also making me accidentally hit Menu some more) for the same deal when it only shuts of the feature instead of letting me adjust it. Again. Judging from the ML-forum. This is unlikely to change. And I guess I’ll just have to live with it and relearn my button-habits for each and every menu item.
I might be sounding negative about those aspects. But overall I love this release! 😀 It’s not perfect… no… and it’ll improve greatly when they enable ML to load automatically during bootup. What are you waiting for?! Get it! (though remember that it probably isn’t covered in warranties).