Must watch video on full frame vs crop cameras. "Full frame look" covered.

[richg101]

Really??

 

...I'm speechless. 

 

I'm not saying I agree completely with his points - and feel he goes about delivery of his points incorrectly, but some of the theories make some sense if the term ISO is used less rigidly than normal.  

[dishe]

You've misunderstood his point.  He is saying that the iso you're seeing displayed on the lcd screen is no longer a tangible figure due to the fact that inside the camera the sensor may be being ragged harder to obtain the same exposure (assuming photosites are similar size).  This is something I had never considered.  The fact that on ex tele mode on the gh2 or on aps-c crop mode on my A7R the noise levels are greater would go some way towards backing up his statements

 

That's always what ISO meant on digital cameras, as I understand it. Every sensor has a sort of "native" exposure level, and ISO applies a gain (either analog or digital or both) to boost the signal or reduce it in some cases. I didn't need Tony to explain that, I learned that when I first starting playing with digital SLRs instead of film cameras.

In order to standardize the values, ISO numbers were created and overseen by the International Organization for Standardization. So ISO 800 on one camera sensor would be approximately the same as ISO 800 on another, despite one applying a different amount of gain to reach that value. 

But it has nothing to do with the overall physical size of the sensor!

[dishe]

The noise being greater when you crop in on your A7R or GH2 is because you are now seeing more noisy pixels since they are larger. They were there before too, but you couldn't see them as the image processor was busy scaling down the image (and taking the noise with it) when they resample it to HD video. Even now, if you take a high ISO HD video and scale it down in post, you'll see less noise, as they get averaged together. 

 

THIS IS NO MERIT TO WHAT THIS GUY IS SAYING!

[Andrew Reid]

The Speed Booster effect has no relation to the size of the sensor. It has a relation to the reduction factor of the Speed Booster and how much the image shrinks, but not the crop of the sensor behind it. The is using a very bad example and technically it is very misleading and confusing.

[noone]

_{I think a lot of what he said is right but he goes too far.      These sort of discussions go on endlessly around the internet though.}  

 

  _{To me, two lenses for different formats (EG 12mm Olympus f2 on micro four thirds against a 24mm Canon FF) give roughly the same picture in terms of angle of view AND shutter speed (if both set to the same aperture and same ISO) BUT will have different depth of field from the same location.}

 

_{It is just as wrong to say they are NOT equivalent if you are only referring to depth of field and ignoring shutter speed.}

 

 

_{So regards the 12-35  2.8 he used to demonstrate against a  24-70 2.8, he is right to say that the 12-35 is NOT a 24-70 FF  equivalent in terms of DEPTH OF FIELD but he is wrong to say it is not a 24-70 FF equivalent in terms of using the same shutter speed ETC.}

 

_{Personally, I like having different sensor sizes for different angle of view  with the same lenses and like to have a different depth of field with different sensors.}

[dishe]

_{I think a lot of what he said is right but he goes too far.      These sort of discussions go on endlessly around the internet though.}  

 

  _{To me, two lenses for different formats (EG 12mm Olympus f2 on micro four thirds against a 24mm Canon FF) give roughly the same picture in terms of angle of view AND shutter speed (if both set to the same aperture and same ISO) BUT will have different depth of field from the same location.}

 

_{It is just as wrong to say they are NOT equivalent if you are only referring to depth of field and ignoring shutter speed.}

 

 

_{So regards the 12-35  2.8 he used to demonstrate against a  24-70 2.8, he is right to say that the 12-35 is NOT a 24-70 FF  equivalent in terms of DEPTH OF FIELD but he is wrong to say it is not a 24-70 FF equivalent in terms of using the same shutter speed ETC.}

 

He said the "aperture changes", which is wrong. It does not. If it did, the exposure would as well. He wants to say that those manufacturers are somehow "lying to you" about the comparable lenses, that it should give comparable F-values as well. But the comparable f values would be far more misleading since they would pertain to light gathering ability!
But that's just one thing he said. Most of his "points" are misguided, especially about how sensor size relates to light gathering ability. In fact, I have more problems with what he described than agreements. Please don't just accept this video at face value.

[Andrew Reid]

When you attach a lens to a crop sensor, the lens remains identical. The image is identical, but for the crop into the image circle.

 

He says the whole photographic system of aperture, ISO, focal length makes no sense. SAY WHAT!?

 

Aperture and focal length relate to the lens only. Why should they apply to a combo of sensor size and lens, as the video suggests it should?

 

A 2x crop factor helps when it comes to figuring out the angle of view. 12 x 2 = 24mm. That's about it.

 

The aperture doesn't change. The exposure doesn't change.

 

Applying crop factor to ISO is a huge mistake. That a larger sensor captures 'more light' ignores the fact that large sensors can still have small pixels due to very high megapixel counts or large gaps in-between pixels.

 

The ISO does not crank up on a smaller sensor because it isn't receiving as much light as a larger sensor.

 

You have to look at the sensor on a per pixel basis. The size of the pixels.

 

Take a 24MP full frame sensor and cut out 16MP from the middle to give you a crop sensor.

 

Look at both images at 1:1...

 

See a difference? No. So why does this guy suggest otherwise?

 

Because what he is neglecting to mention is he's not looking at the images 1:1, but downsampling the 24MP to 16MP, then comparing to a 16MP crop sensor. The downsampling of course makes for a cleaner image.

 

This is also where DXOMark goes wrong very often. They include the downsampling in their calculations too.

 

So a 36MP sensor would do very well in low light even if it was noisier on a per pixel basis. Look at their low light results for the 5D Mark III vs D800. We know the 5D Mark III has better high ISO than the D800 yet DXOMark somehow haven't managed to show this in their scores.

 

What matters is pixel size, pixel architecture and readout noise... Not sensor size.

 

The Blackmagic Production Camera is absolute proof that this guy's theory fails the reality test.

[KarimNassar]

I appreciate the discussion in here thank you all, but I still disagree with you.

 

Dishe I believe you are wrong and  I feel you are contradicting yourself:

 

the 1st and 3rd have the same intensity of light hitting the sensor at a time

But the middle one does not- it is brighter than the others because in reducing the image, the photons are also condensed (less scattered) which results in more intense light. 

 

If the light is more intense when condensed, then it is obligatory and by definition less intense when not condensed.

 

You are acknowledging that the speedbooster is condensing light resulting in more intense output, yet you do not want to acknowledge that a less condensed output results in less intensity. 

 

That is simply not possible.

 

I believe your magnifying glass analogy might be the root of the disagreement.

 

Unlike a magnifying glass, the speedbooster is placed behind the lens.

 

Which means:

 

- it can only focus the light the lens is letting through, it will never add light.

 

- at f1.8 a full frame lens will always, regardless of the sensor size, let the same amount of light through

 

- a full frame sensor takes full advantage of this f1.8 light output because the sensor area covers more of the light beam (see my previously posted graphic).

 

- a crop sensor is missing a lot of light coming in from this f1.8 opening as lots of it lands outside of the crop sensor (see my previously posted graphic).

 

- this is why a speedbooster, while increasing the amount of light hitting a crop sensor resulting in a more exposed crop sensor, is not actually increasing the amount of light let through the 1.8 lens, only making up for the loss that was previously missing and hitting outside of the crop sensor.

 

Which is why ; without a speedbooster, you must multiply not only the focal length to get the field of view equivalence, but also the aperture to get the equivalence.

 

And with a speedbooster, which is only making up for the loss, not adding more light, a f1.8 full frame lens remains a f1.8 lens.

 

 

This is how I see it.

 

 

 

 

[jasondhsd]

I don't have time to watch the video but f2.8 is f2.8 in regards to the amount of light hitting the sensor whether it's hitting a sensor that's 5 inches or a sensor that's 1cm the same amount of light is hitting the sensor.  Now pixel size is a factor an APC-C sensor is about 30% larger then a m43 sensor but it also has about 30% more pixels per square inch (or whatever unit of measurement) so a micro43 sensor at 16 megapixels has slight smaller pixel then a 20.1mp canon and slightly larger pixels compared to a Nikon at 24mp.   But take something like a full-frame sensor like the MD MkIII at 24mp, compared to the apc-c nikon it's the same amount of pixels spread out over a larger area. And then you have the Nikon 4ds full frame with 16mp those pixels are HUGE and you're able to boost the gain a lot higher compared to m43 and apc-c where the sensor is more densely packed, which basically means full frame can take the same amount of light and use it better at higher ISO.  But at lower ISO say 3200 or 1600 and less set all the cameras m43, apc-c, full frame to same frame rate, iso, and shutter speed and the photos will all be exposed exactly the same.    Now depth of field is a completely different matter but has nothing to do with light transmission. tl, dr: the same amount of light is being let in through the hole regardless of the sensor size. 

 

As for speedbooster,what that's doing is taking the f-whatever lets say f2.8 taking that light and focusing it on a smaller area or smaller sensor....like a magnifying glass focusing the sun, so f2.8 becomes f2. 

[jcs]

Haven't watched the whole video; he makes a few mistakes, however his math makes sense so far and the images and settings match the math predictions. So far, he is generally correct. Math followed up with experiments and examples that match the math is a strong argument. Any challenges should be made to his math and the experimental results.

[jasondhsd]

From metabones:

 

 

 

 

• Does the Speed Booster® increase only T-stop of the lens leaving F-stop unchanged?

  This is one of the common misconceptions about the Speed Booster®. However there is a contradiction right within that assertion since T-stop cannot be any faster than F-stop. It is not possible to experimentally observe a T-stop increase unless the lens has a corresponding F-stop increase.

  The logic of the allegation is that since the depth-of-field of the lens does not change, therefore neither does the F-stop of the lens (untrue). What had never been under any dispute was that the T-stop of the lens did increase, as could be seen with the increased exposure in the resulting footage or photograph.

  Before we clear up this misconception, let's find out why there is so much confusion in the first place. Focal length, maximum aperture and depth-of-field are physical quantities that are independent of sensor size. These quantities do not care whether a full frame, APS-C or m4/3 sensor sits behind the lens.

  Speed Booster® makes the focal length 0.7x shorter. F-number is simply focal length divided by entrance pupil diameter. Since the former reduces by a factor of 0.7x but the latter remains the same, F-number also becomes 0.7x smaller, or one stop faster. The F-stop increase is real.

  Note that a 35/1.0 lens (from 50/1.4 + Speed Booster®) will always have shallower depth-of-field than a straight 35/1.4 lens at any given distance.

[noone]

I don't have time to watch the video but f2.8 is f2.8 in regards to the amount of light hitting the sensor whether it's hitting a sensor that's 5 inches or a sensor that's 1cm the same amount of light is hitting the sensor.  Now pixel size is a factor an APC-C sensor is about 30% larger then a m43 sensor but it also has about 30% more pixels per square inch (or whatever unit of measurement) so a micro43 sensor at 16 megapixels has slight smaller pixel then a 20.1mp canon and slightly larger pixels compared to a Nikon at 24mp.   But take something like a full-frame sensor like the MD MkIII at 24mp, compared to the apc-c nikon it's the same amount of pixels spread out over a larger area. And then you have the Nikon 4ds full frame with 16mp those pixels are HUGE and you're able to boost the gain a lot higher compared to m43 and apc-c where the sensor is more densely packed, which basically means full frame can take the same amount of light and use it better at higher ISO.  But at lower ISO say 3200 or 1600 and less set all the cameras m43, apc-c, full frame to same frame rate, iso, and shutter speed and the photos will all be exposed exactly the same.    Now depth of field is a completely different matter but has nothing to do with light transmission. tl, dr: the same amount of light is being let in through the hole regardless of the sensor size. 

 

As for speedbooster,what that's doing is taking the f-whatever lets say f2.8 taking that light and focusing it on a smaller area or smaller sensor....like a magnifying glass focusing the sun, so f2.8 becomes f2. 

No, there IS more light hitting a larger sensor if both have a 2.8 lens but not more light per area.    The light hits all of the sensor at the same time so with a larger area there MUST be more total light hitting the sensor.  The SAME amount of light hits the same size area though.

[jasondhsd]

No, there IS more light hitting a larger sensor if both have a 2.8 lens but not more light per area.    The light hits all of the sensor at the same time so with a larger area there MUST be more total light hitting the sensor.  The SAME amount of light hits the same size area though.

 

A lens at f2.8 is letting in the same amount of light whether the light is hitting something an inch or a millimeter its still the same amount of light. A larger sensor sitting behind the lenses doesn't get more light then a crop sensor or a m43 sensor or any size sensor sitting behind the lens. 

[dishe]

I don't even know where to begin. The science and real-world practical application both favor the explanation Andrew and I have been trying to give you. Forget about the physics behind it (which totally back up our explanation), I actually OWN speedboosters (focal reducers technically), crop sensors and full frame sensors. And I can tell you first hand what it does, or does not do. 

 

We're not wrong. 

But because I'm a sucker for a good internet argument, I'll play along a little more. 
 

 

Dishe I believe you are wrong and  I feel you are contradicting yourself:
 
 
If the light is more intense when condensed, then it is obligatory and by definition less intense when not condensed.

 
Correct. On the full frame camera without a speedbooster, it is not condensed. On the crop sensor without a speedbooster, it is... wait for it... ALSO not condensed. Hurray, we agree!
 

You are acknowledging that the speedbooster is condensing light resulting in more intense output, yet you do not want to acknowledge that a less condensed output results in less intensity.

 
Um- who say what now? Yes, if the image circle is larger, the light is less intense. We're still in agreement! ...I think?
 

I believe your magnifying glass analogy might be the root of the disagreement.
 
Unlike a magnifying glass, the speedbooster is placed behind the lens.

Actually, that's not unlike a magnifying glass at all. Or a pair of glasses. Or any optical element at all. You might even say it is very similar, except facing the other direction.
 

Which means:
 
- it can only focus the light the lens is letting through, it will never add light.

No lens can "add light" that doesn't exist. However, it can intensify it by condensing it. Just like bringing a flashlight closer to the wall makes the circle smaller and brighter, or the magnifying glass makes the sun light brighter and hotter. Is it making adding to the sunlight? Is the flashlight producing more photons? 
No. They are just not as scattered, and therefore brighter and more intense per surface area. The reason I can't take my projector and point it at the sky to get a giant movie on a cloud is because the image stretches out so far that almost all the illumination is lost. But conversely, gather all the light into a single point, and it becomes intensely bright. That's actually how a laser beam works. Hurray for science!

 

- at f1.8 a full frame lens will always, regardless of the sensor size, let the same amount of light through

True. But take that circle of light, pretend it is coming out of a flashlight. Make it smaller by optically reducing it. That smaller projection of light is brighter as a result.  
 

- a full frame sensor takes full advantage of this f1.8 light output because the sensor area covers more of the light beam (see my previously posted graphic).

Light output is the same. The part of the image that hits the sensor is the same intensity as the light that misses it. Those pixels in the middle of a full frame camera get exposed the same amount as a cropped one. I can't explain this any better than that. 
 

- a crop sensor is missing a lot of light coming in from this f1.8 opening as lots of it lands outside of the crop sensor (see my previously posted graphic).

True. Which means diddly squat for the pixels that DID get exposed. You're missing the point here- you're looking at the sensor size as a whole, which is silly. A digital image is merely a grid of single pixels, each one unaware of the others around it. The grid of a sensor is made up of photosites, each one responsible for roughly a pixel (oversimplification, but works for the purpose of this discussion). When a photosite is shown that bit of light it can see from the lens, it sends an electrical signal to the image processor which turns that information into a value of light. This happens regardless of how many OTHER photosites there are around it, however a larger one can collect more photons and render a more accurate color and exposure. That is why larger photosites (such as those usually found on larger sensors of the same megapixel count) can have less noise than smaller ones. But this isn't a function of the size of the sensor. Its a function of the density of photosites on that sensor. While a larger sensor has more room for big photosites, it isn't necessarily always the case (especially with 36Mpix full frame sensors!).

The point is, the light that falls outside the sensor makes no impact on the light that exposes in the middle. On a full frame sensor, the light that falls on the middle photosites are the same intensity as those on a crop. It really makes no difference and it really is as simple as that. 
 

- this is why a speedbooster, while increasing the amount of light hitting a crop sensor resulting in a more exposed crop sensor, is not actually increasing the amount of light let through the 1.8 lens, only making up for the loss that was previously missing and hitting outside of the crop sensor.

No. It is reclaiming the area that fell outside the sensor previously, but that is in regards to the field of view only. A happy side effect of condensing the light, however, is that it also becomes more intense. In fact, the math goes something like this:

A full frame 50mm 1.8 on a Sony NEX speedbooster gives you the field of view of a full frame 50mm and DOF of a 1.8, but the light gathering ability of a F/1.2
 

Which is why ; without a speedbooster, you must multiply not only the focal length to get the field of view equivalence, but also the aperture to get the equivalence.
 
And with a speedbooster, which is only making up for the loss, not adding more light, a f1.8 full frame lens remains a f1.8 lens.
 
 
This is how I see it.

 
I understand that's how you see it. Unfortunately, how you feel about it has very little impact on the science and actual real-world usage of the equipment. 

Case-in-point. Someone took a Sony A7 full frame sensor and stuck a NEX speedbooster on it for kicks. The image now takes up only the middle part of the frame and vignettes something fierce (edges are all black), and is brighter than it is without the speedbooster. That's just how this works. 

 

 

Haven't watched the whole video; he makes a few mistakes, however his math makes sense so far and the images and settings match the math predictions. So far, he is generally correct. Math followed up with experiments and examples that match the math is a strong argument. Any challenges should be made to his math and the experimental results.

 

Care to elaborate as to which mathematics make sense to you?

The smaller sensor has more noise because the photosites are smaller and therefore less light sensitive, the DOF changes because you need to place your subject farther away from the camera (DOF being a calculation of Aperture to distance). The things he is showing work for a completely different reason than he is saying they do. The larger sensor does not gather more light on account of being larger, the aperture does NOT change, following his math does NOT add up!

[noone]

A lens at f2.8 is letting in the same amount of light whether the light is hitting something an inch or a millimeter its still the same amount of light. A larger sensor sitting behind the lenses doesn't get more light then a crop sensor or a m43 sensor or any size sensor sitting behind the lens. 

Ok yes I worded that wrong.    The same amount of light goes through the same size aperture.     The 24-70 2.8 lens gets more total light than the 12-35 2.8 because of the larger aperture (in terms of area) but because it hits a sensor with a larger area they get the same light per sensor area.    

 

EDIT    These discussions go around and around and sometimes I get lost as to which "side" I am on.

I think a lot of what is in that video is right but he does go too far.

[Andrew Reid]

- at f1.8 a full frame lens will always, regardless of the sensor size, let the same amount of light through

- a full frame sensor takes full advantage of this f1.8 light output because the sensor area covers more of the light beam (see my previously posted graphic).

- a crop sensor is missing a lot of light coming in from this f1.8 opening as lots of it lands outside of the crop sensor (see my previously posted graphic).

 

Dude this is wrong.

 

A full frame sensor will only take advantage over the larger image circle projected by the lens if the pixel architecture is sensitive enough. If you have small pixels with big gaps in-between or too many pixels and too small pixels then your theory will snap like a brittle stick in the dog mouth of reality.

 

For example a smaller crop sensor with larger pixels than a larger full frame sensor will capture more light.

 

Per-pixel quality matters the most.

 

Therefore you cannot simplify this down to simply 'a larger sensor captures more light' like the idiot in the video has got everyone doing. Ridiculous AND annoying at the same time.

[jcs]

Andrew- let's start with his math: where are the errors?

He shows example images which match his math: what's wrong with the images?

[Andrew Reid]

Andrew- let's start with his math: where are the errors?

He shows example images which match his math: what's wrong with the images?

 

It's not about math or his example images, it really isn't. He uses very little maths and it is mainly his other examples which aren't backed up by any maths at all that are so wrong. What he is saying about applying crop factor to ISO is just plain wrong. It is like calculating the speed of an aircraft based on how windy it is outside, and ignore all the other things. The examples he's grasping at are bad. The reality is very different to what he describes in the video.

 

The intensity of light hitting the pixels and the design of those pixels are what matters when it comes to sensitivity, ISO and aprture, NOT the area of capture or crop factor of the sensor!

 

This is infuriatingly obvious stuff!

[Andrew Reid]

Say if we follow his logic and apply crop factor to ISO...

 

5D is an old camera, can barely do clean ISO 800 but hey... BIG SENSOR!

 

GH4 is a new camera, does clean ISO 800 on a 2x crop sensor.

 

By this logic 800 is actually 2x... so as noisy as ISO 1600 on the 5D Mark 1's full frame sensor from years ago...

 

Nah doesn't work does it!?

[Andrew Reid]

I am a big fan of the full frame look and large sensors but facts are facts guys.