Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Magic lantern... Dude.

David Mullen ASC said:
Is it a sensor crop, other than the vertical crop to 2.40? Sounds like it's the sensor signal scaled from 5616 to 1920 pixels across using whatever line skipping / pixel binning, etc. method the camera normally uses to get HD video off of the sensor, otherwise no one would get that shallow focus full-frame look in their HD video. It seems to me that they are somehow capturing the raw signal somewhere in mid-processing, after the line skipping / pixel binning but before the conversion to RGB. There may be some cropping but I don't think it's being cropped from 5616 pixels to 1920 pixels, that would be pretty severe.
Click to expand...

You could be onto something in terms of it being a bit of both. Heres a list of the pre-set resolutions:

static int resolution_presets_x[] = { 640, 720, 960, 1280, 1320, 1920, 2048, 2560, 2880, 3592 };
static int resolution_presets_y[] = { 320, 360, 480, 540, 720, 840, 960, 1080, 1152, 1280, 1320 };

I notice that these top out at 3592 on the x axis.

Our friendly magic lantern developer goes onto say:

or add your own custom resolutions (just put the numbers there).

The high resolutions only work in 5x zoom mode, where you get 1:1 crop from the sensor.

Right now, ML will do a center crop, but I'd like to try to implement a smooth digital panning (sort of a software dolly). Shifting the crop window by 8 pixels at every frame should be easy, and that would result in a 8-second horizontal panning of the 1920x1080 window from sensor center to the edge: (3592-1920) / 8 / 24fps = 8.7 seconds.


Which would very much seem to imply it is quite a big crop and that yes, they are going to lose much of the shallow depth of field look when recording in RAW. I guess you could continue to use .h264 for the shallow depth of field shots tho. On the upside, given that they are starting with a full frame sensor, the crop is still going to give a bigger sensor area than might be expected in other situations. (IYSWIM, I explained that bit badly but I'm hoping you get what I'm getting at because It's late at night here!)

I notice tho that they really do seem to be talking about a maximum of 3592 (sensor edge) which is different to the 5616 pixels you speak of.

Dunno.

Theres definitely a sensor crop involved tho, both here and mentioned above.

Also I'm assuming the 1920x1080 crop is 16:9.
It was a different breakthrough earlier today where they had continuous recording going for a 2.40:1 aspect ratio.

You are definitely asking good questions tho David. If it's a big sensor crop there is also likely to be the same kind of issue you have on the Red One working in 2k as well where you end up having extra lenses around for 2k so you can get wider angle shots in that mode!

Obviously the larger the sensor crop the more telephoto the lenses will appear to be.

Freya
 
Freya said:
Which would very much seem to imply it is quite a big crop and that yes, they are going to lose much of the shallow depth of field look when recording in RAW. I guess you could continue to use .h264 for the shallow depth of field shots tho.
Click to expand...

Cropping sensor has nothing to do with shallow DoF. There seems to be some FUD spread about this all over the web. The DoF depends on the lens design and "speed" (how wide open the aperture is at the time of capture). Cropping sensor is just what it says it is - cropping. You simply only record a smaller portion of the full frame image - which has no effect on DoF...

Peter
 
Peter Majtan said:
Cropping sensor has nothing to do with shallow DoF. There seems to be some FUD spread about this all over the web. The DoF depends on the lens design and "speed" (how wide open the aperture is at the time of capture). Cropping sensor is just what it says it is - cropping. You simply only record a smaller portion of the full frame image - which has no effect on DoF...

Peter
Click to expand...

But that does indeed affect how shallow DOF "feels", even though you are technically correct as to measurements...

try shooting a 50@ 1.3 @ FF 35, @ S35 and at S 16/2k HD.

If you adjust for framing to get the same frame with the different sensor-crops, (by movig the camera), the apparent DOF will change a lot.

But technically, you are of course right.

If you just crop the sensor and thus just change FOV, the DOF does not change.
 
If you crop the sensor you end up using shorter lenses to achieve the same field of view and getting the depth of field characteristics of a smaller format, which is generally deeper than a larger format. This isn't FUD, it's just basic photography.
 
there may be some vertical cropping but i'm almost positive this is a scale and it's not the regular video mojo... like i think you get .DNG's, like a really fast timelapse. it's just like doing a burst at speed, but it keeps going because the processing/saving is limited in size.
 
Based on one online DOF calculator:

Let's say that you are using a Canon 7D with an APS-C sensor with a 50mm at f/2.8 focused at 5'... the DOF range is 4.85' to 5.16' (.32' total).

You switch to a Canon 5D with a FF35 sensor and switch to a 75mm to maintain a similar field of view, also at f/2.8 focused at 5', then the DOF range is 4.89' to 5.11' (.22' total).

I know this is basic stuff here but smaller sensor areas, whether due to cropping or using a physically smaller sensor, generally leads to deeper focus photography after you've switched to shorter focal lengths to maintain the same field of view at the same shooting stop and distance to subject. So this is not FUD, sensor size/area has an affect on depth of field characteristics, anyone who has switched from a 1/3" sensor camera to a FF35 sensor camera will experience this.
 
This is BIG news.
And trouble for a lot of other camera makers in the industry (including Canon themselves).
 
David - no questions about the implications / consequences...

The simple point was that cropping sensor "alone" does absolutely nothing to DoF.
The same lens on the same sensor at same aperture will provide exactly the same DoF regardless of how much you crop the sensor.
Cropping sensor is equivalent to cropping the resulting picture in post.
Cropping sensor changes the FOV. Which in return forces one to use wider lenses - which do have shallower DoF.
But it is the change of lenses (which does change the optical path) that causes the DoF, not the cropping of the sensor...

Was just trying to make that clear... (see the OP's sentence that I have quoted and the implication it might have on people without experience - whom in return spread FUD based on such a posts)...

P
 
okay, taking a further look at it i'm 99% sure we're not talking about a full-on sensor crop (like we're used to on red, but don't quote me), though there are some options for vertical/aspect cropping. it is taking the images from the live view feed, which apparently at this point maxes at 3.5k (not at speed) as a 422 YUV post-converted to DNG (14-bit, i think)... i'm sure this will stabilize a little bit as it's all pre-alpha, but i see the hdmi out (magic lantern or canon's) being a 'safer' route for capturing anything important. lots of weird issues and artifacts that won't go away regardless. probably a better bet to go bm or canon cine in this price range... but i like how these dudes are pushin' code.
 
Peter Majtan said:
David - no questions about the implications / consequences...

The simple point was that cropping sensor "alone" does absolutely nothing to DoF.
The same lens on the same sensor at same aperture will provide exactly the same DoF regardless of how much you crop the sensor.
Cropping sensor is equivalent to cropping the resulting picture in post.
Cropping sensor changes the FOV. Which in return forces one to use wider lenses - which do have shallower DoF.
But it is the change of lenses (which does change the optical path) that causes the DoF, not the cropping of the sensor...


Was just trying to make that clear... (see the OP's sentence that I have quoted and the implication it might have on people without experience - whom in return spread FUD based on such a posts)...

P
Click to expand...

Interesting. So Super16 has the same level of depth of field as Super 35mm and a Full Frame camera gives the same amount of depth of field as a Red One then, it's just a smaller image area.

A 50mm lens will thus give the same level of depth of field on any camera, assuming the aperture remains consistent.

Handy to know!

Does this mean that those DOF adaptors people used to use were just a bit of a con?

Freya
 
paulherrin said:
okay, taking a further look at it i'm 99% sure we're not talking about a full-on sensor crop (like we're used to on red, but don't quote me), though there are some options for vertical/aspect cropping. it is taking the images from the live view feed, which apparently at this point maxes at 3.5k (not at speed) as a 422 YUV post-converted to DNG (14-bit, i think)... i'm sure this will stabilize a little bit as it's all pre-alpha, but i see the hdmi out (magic lantern or canon's) being a 'safer' route for capturing anything important. lots of weird issues and artifacts that won't go away regardless. probably a better bet to go bm or canon cine in this price range... but i like how these dudes are pushin' code.
Click to expand...

I'm preety sure it's pre live view, as they have been trying out turning the live view off to try and increase the speed of things. I think it's also ahead of the 422 YUV, as until the RAW breakthrough they were experimenting wtih recording the YUV and if that was higher up the chain, you would think they would continue with that rather than switching to trying to record the RAW.

Of course that still doesn't mean that this isn't a combination of sensor crop and something else.

Freya
 
peter, i think everyone here (that will ever need to) is able to make the distinction between depth of field as a technical and a visual term.

for anyone that's confused, depth of field technically refers to the area of focus as distinguished from what's out of focus (hence, shallow/deep) - although such measurements can get a little 'blurry' and subjective, which is why relative/visual terms are more typically used unless you're doing highly technical work (in which case you should seek information elsewhere).

anyways, i'm sure there are better things to get done than semantic debates on reduser about dof-fud. :)
 
Freya said:
I'm preety sure it's pre live view, as they have been trying out turning the live view off to try and increase the speed of things. I think it's also ahead of the 422 YUV, as until the RAW breakthrough they were experimenting wtih recording the YUV and if that was higher up the chain, you would think they would continue with that rather than switching to trying to record the RAW.

Of course that still doesn't mean that this isn't a combination of sensor crop and something else.

Freya
Click to expand...

yeah maybe, i was referring to the live view feed - i'm assuming there is a break in the chain where it either goes to live view/hdmi or compression, but i'm not qualified to answer that. so it's hard to say exactly what 'raw' is in this case other than what it's not - h264 iframes.
 
Freya said:
Interesting. So Super16 has the same level of depth of field as Super 35mm and a Full Frame camera gives the same amount of depth of field as a Red One then, it's just a smaller image area.

A 50mm lens will thus give the same level of depth of field on any camera, assuming the aperture remains consistent.

Handy to know!

Does this mean that those DOF adaptors people used to use were just a bit of a con?

Freya
Click to expand...

Not at all.

The component Peter overlooks is that size of focal plane is a huge factor of the perception of relative DOF.

The area which is measurable in focus at a given focal length at a given aperture is identical, but as the focal plane changes, the relative DOF is higher (more of the resulting image is in focvus) if you frame the same shot with a wider lense or pull back to get the same FOV with an identical lens, at the same aparture.

Makes sense?
 
Not a joke at all!

When changing from a 15mm to a 300mm, "Perspective changes dramatically, but not depth of field". Technically, that's true, and visually, it's also true. I've read your post, won't bother with Wiki and hell, bantering never taught anyone anything, but neither did accepting someone else's opinion without reviewing it further. That post at LL makes perfect sense to me and honestly thought it would be interesting for those who are so sure "depth of field is narrower with longer lenses", because after reading that article, I don't think that statement is accurate. Perception is one thing, but it's fun to know what is really happening, wouldn't you think?
 
Gunleik Groven said:
Not at all.

The component Peter overlooks is that size of focal plane is a huge factor of the perception of relative DOF.

The area which is measurable in focus at a given focal length at a given aperture is identical, but as the focal plane changes, the relative DOF is higher (more of the resulting image is in focvus) if you frame the same shot with a wider lense or pull back to get the same FOV with an identical lens, at the same aparture.

Makes sense?
Click to expand...

That's perhaps because I was not talking and subjective elements such as "perception" and "relative".
The point I am making is that "cropping" sensor DOES NOT change DoF.
Take your Epic / Scarlet / RED ONE and shoot the same scene in various resolutions. Then take the highest resolution frame and crop it in Photoshop to match the sizes of the smaller resolutions. You will end up with IDENTICAL images. Period.
It is the consequences of the change in FoV (not DoF) that forces one to change the lens and/or camera position - which then have the direct effect on DoF...

Guns do not kill people. People kill people...

:sifone: Peter
 
Lets talk about the raw video here and leave the two pages of dof chat elsewhere.

It's a pretty astounding development, i agree that rolling shutter does seem to be the big question.
 
You must log in or register to reply here.
Back
Top