Nasty color bleed on pushed A7III RAW (compressed, silent shutter)

Started 7 months ago | Discussions thread
PhozoKozmos
PhozoKozmos Contributing Member • Posts: 792
Re: SONY steep short+wide optical projection to sensor is cause (not file format related)

JimKasson wrote:

PhozoKozmos wrote:

JimKasson wrote:

PhozoKozmos wrote:

JimKasson wrote:

PhozoKozmos wrote:

JimKasson wrote:

PhozoKozmos wrote:

JimKasson wrote:

PhozoKozmos wrote:

SONY steep short+wide optical projection to sensor is cause (not file format related)

anything in foreground out-of-focus will have an expanded blurred ghost

but if it is projected at a very wide-angle onto a very close sensor surface, more of the ghost can be spread farther across more adjacent pixels and pick up the extra light

unlikely, file format related, as ooc (fine only) Sony jpeg paired with Sony RAW will exhibit same effect (regardless of compression or not, and mechanical shutter or not)

This effect is not FFD related here, for even ray-angle related (except the calculation of CoC being focal-plane focus error over f-stop). It is entirely explained by the out-of-focus point spread functions of defocused rays convolved with the objects in front of the lens.

oof point spread is normal, yes

however, i suspect the steep ray-angle may be susceptible in exacerbating a wider spread than usual

Nope. OOF PSF is a function of focus plane error and f stop only, if there are no internal lens occlusions.

You need to address the point above for your argument to have a chance of being credible, I think.

i'm not attributing oof psf to lens flaws, per se

but rather to specific lenses with very short-wide projection geometry of needing to cover a wide area (FF) in a very short throw

much like "short throw" light projectors, situated very close to projection surface

the closer, the more difficult to design the optics

it includes exacerbated chromatic aberration spread

You talking LaCA? If so, that is not true, in my experience.

you may simply have chosen better lenses, so you will not encounter it

short ffd only applies for lenses with a short-but-wide lens-sensor dimension LSFD

The FFD doesn't change when you change lenses.

agree. body-flange-sensor dimension (BFSD) never changes, it is fixed 18mm for A7III; but this has nothing to do with lens-sensor focal dimension (LSFD)

not long LSFD

not exceptionally short-but-shallow LSFD (like lenses with 16.7mm back focus, where lens optic rear opening is huge, so projection is shallower)

IMHO, this is a complete red herring.

e.g.

fujifilm GFX50S have primes like the new 250mm f4, with 57.1mm (minimum) LSFD, even though its body-flange-sensor dimension is a fixed 26.7mm. other fujifilm lenses could have as little as 16.7mm LSFD (minimum) according to fujifilm GFX50S specs

LSFD varies with specific lens

Define LSFD, and tell me how it is different from the location of the exit pupil.

Define "exit pupil".

The exit pupil of a lens is the image of its aperture stop as seen from the back of the lens. Relevant parameters are the size of the pupil, its shape, and the distance from the focal plane.

https://en.wikipedia.org/wiki/Exit_pupil

ok. got it.

keyword is "image of aperture"

for a poor lens, with say high LaCA:

  • that "image of aperture" will have yellow fringing on the nearer aperture edge
  • that "image of aperture" will have blue fringing on the far (opposite) aperture edge

when either are projected onto the focal plane (and sensor stack):

  • the nearer yellow fringe will project at a shallower angle along with unfringed colors, and not stand apart as strongly as blue fringing, on image plane (including sensor stack)
  • the farther blue fringing will project at very wide steeper angle more separated from unfringed colors, on image plane (including sensor stack)

It may coincide with LSFD, or may be unrelated.

LSFD = Lens (Rear of last Optic) to Sensor Plane Focal Dimension (distance)

It turns out they are not the same, then. But LSFD is irrelevant. The last optic in an a7x system is the sensor stack, right? It is very close to the focal plane.

wrong:

  • the LAST optic of the interchangeable lens, is at the REAR of the lens
  • the FIRST optic of an a7x/a9 system is the sensor stack, very close to the focal plane

But that's got nothing to do with ray angles.

ray angles are shallow, nearly parallel to lens axis within the sensor stack, projecting onto the focal plane

however, ray angles will vary far off from lens axis, by lenses, by how far the last optic in the rear of the len is to the first optic (sensor stack) acting like a projection screen over the focal plane

The exit pupil does.

i think we are refering to the same rear optic of the lens, here. (=exit pupil)

The exit pupil can be at the same location as the last non-stack element in the lens.

yes. for wider-angle lenses, provided the rear lens diameter optic is especially large, relative to sensor size, then LSFD is "short-large-shallow"

this projection geometry would match longer LSFD

again: not for those that are "short-wide-angle (steep)" LSFD (more common on cheaper lenses)

The exit pupil can be closer to the sensor than the last non-stack element in the lens. The exit pupil can be farther away from the sensor than the last non-stack element in the lens. It all depends on the way the lens is designed.

Sony's new (pending) tele prime has an exceptionally long LSFD, so it will have a normal (good) oof psf (typical for telelenses) and lower LaCA (predictable, as it mimic longer LSFD of dSLRs)

As far as I'm concerned, any relationship between exit pupil location and LaCA is totally unproven.

I make no mention of "exit pupil"

But that is the important thing for ray angle.

we differ in terminology. so, i agree.

"It's way more than terminology. You said: " LSFD = Lens (Rear of last Optic) to Sensor Plane Focal Dimension (distance)". OK, so you don't count the sensor stack as an optic, even though it is one. So you're talking about the distance between the last piece of glass in the lens and the sensor. But that does not determine ray angles. The exit pupil location does.

no. i didn't say rear of lens is exclusively last optic before sensor plane.

here, graphic version:

LENS + (lens-span-to-sensor) + SENSOR

[1st-optic+(middle-optics)+last-optic] |<=LSFD=>| [1st-optic(stack)+sensor-plane]

missing: body-flange-sensor dimension (irrelevant)

however, not all lenses enjoy longer LSFD, nor short-but-shallow LSFD

however, any absence of CA at the center of focal plane of sensor

does not preclude LaCA off center (a given)

however, a long LSFD, will have LaCA of a specific ANGULAR spread, say 1" second (for illustration only)

a shorter LSFD, with the same LaCA Angular spread, of 1", will cover more area (=wider) off center

better lenses, with long LSFD, will not suffer as bad as lenses with shorter LSFD

i don't doubt you select the best lenses for your mirrorless cameras, but i'm certain they will all have quite long LSFD.

others may not be so fortunate, choosing shorter LSFD lenses (worse for WA, than for tele)

A can't find anything in those last 7 paragraphs that makes sense to me. Why is LSFD significant at all? It's got nothing to do with ray angles.

it defines ray angles, more than body-flange to sensor dimensions alone.

No, it doesn't. I think I explained why. We need to get this straightened out before we get into the effect of different angles of incidence, about which I also disagree.

Jim

-- hide signature --

there are two ways ray angles will be projected from a lens to the sensor

1. nearly parallel to lens axis (can use long or short LSFD), nearly perpendicular to sensor. here, the sensor is treated like "another lens element"

  • long LSFD is typical for good longer FL lenses or good short FL
  • (ultra-short LSFD is typical for good "wide" focus-reducers, or fixed zoom dcams)

2. convergent rays, at a steeper angle from lens axis

  • could be a mild angle of incidence projection onto image plane (=good), typical of longer LSFD
  • but could be a very steep angle of incidence projection onto image plane (=bad); too steep, spells trouble, typical of too short LSFD (without parallel rays, in case 1, above)
-- hide signature --

pzk
phos.zoe.kosmos (light.life.world)
phos-kosmos (light of the world)
phos-zoe (light of life)
photo-graph-them-all!

 PhozoKozmos's gear list:PhozoKozmos's gear list
Canon PowerShot SX50 HS Canon EOS 5D Mark II Canon EF 24mm f/1.4L II USM Canon TS-E 24mm f/3.5L II Venus Laowa 12mm F2.8 Zero-D
Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow