Real vs market ISO - page 10 - General Discussion and Assistance - CHDK Forum
« previous next »
Pages: 1 ... 5 6 7 8 9 [10] 11 12   Go Down

Real vs market ISO

  • 110 Replies
  • 37150 Views
*

Offline philmoz

  • *****
  • 3450
    • Photos
Re: Real vs market ISO
« Reply #90 on: 28 / July / 2013, 17:39:12 »
Advertisements
Quote from: reyalp on 28 / July / 2013, 17:21:24
Quote from: philmoz on 28 / July / 2013, 17:02:59
Assuming the hardware even cares, is this worth worrying about?

As I said, there's no discernible difference between a CHDK ISO 80 image and a Canon ISO 80 difference.
I don't understand what the problem is. What are the propcase values at ISO 80?


That's all I want to know.

As I said before, none of this really matters for 1.2, since the special case is commented. Having the correct values will allow us make it more accurate for 1.3. Maybe it doesn't make any difference, but we will have to fix the rest of the override code to find out. Since you have the cameras and the information, it would be helpful to know the actual values.

edit:
OK, from the conversion factor, I see it's 370/449 like other modern cameras, so it should not undef the low values.

The point I was trying to make is whether this special case code for ISO 80 is needed at all, for a difference that can't be seen in the end result.

Phil.
Logged
CHDK ports:
  sx30is (1.00c, 1.00h, 1.00l, 1.00n & 1.00p)
  g12 (1.00c, 1.00e, 1.00f & 1.00g)
  sx130is (1.01d & 1.01f)
  ixus310hs (1.00a & 1.01a)
  sx40hs (1.00d, 1.00g & 1.00i)
  g1x (1.00e, 1.00f & 1.00g)
  g5x (1.00c, 1.01a, 1.01b)
  g7x2 (1.01a, 1.01b, 1.10b)
*

Offline reyalp

  • ******
  • 14128
Re: Real vs market ISO
« Reply #91 on: 28 / July / 2013, 17:58:37 »
Quote from: philmoz on 28 / July / 2013, 17:39:12
The point I was trying to make is whether this special case code for ISO 80 is needed at all, for a difference that can't be seen in the end result.
Possibly not, but looking at images with the current override code is not a good indicator, since the ISO->apex96 conversions have other errors that make the resulting override low. In the current code, without the special case, setting an override of 80 ends up with APEX96 values equivalent to something like 74. You can see this if you show ISO in the misc values (though this may vary due to other issues)
Logged
Don't forget what the H stands for.
*

Offline philmoz

  • *****
  • 3450
    • Photos
Re: Real vs market ISO
« Reply #92 on: 28 / July / 2013, 19:49:35 »
Quote from: reyalp on 28 / July / 2013, 17:58:37
Quote from: philmoz on 28 / July / 2013, 17:39:12
The point I was trying to make is whether this special case code for ISO 80 is needed at all, for a difference that can't be seen in the end result.
Possibly not, but looking at images with the current override code is not a good indicator, since the ISO->apex96 conversions have other errors that make the resulting override low. In the current code, without the special case, setting an override of 80 ends up with APEX96 values equivalent to something like 74. You can see this if you show ISO in the misc values (though this may vary due to other issues)

The Misc values for ISO are showing incorrect value for every CHDK ISO override, not just 80.

For example if I select Canon ISO 400 the Misc values shows I-R = 261, I-M = 400.
If I then use the CHDK override to select ISO 400 I see I-R = 241, I-M 369.

I would expect to see a visible difference in this case between the Canon and CHDK settings; but I don't.

This would seem to indicate that the Misc values display is broken.

Phil.
Logged
CHDK ports:
  sx30is (1.00c, 1.00h, 1.00l, 1.00n & 1.00p)
  g12 (1.00c, 1.00e, 1.00f & 1.00g)
  sx130is (1.01d & 1.01f)
  ixus310hs (1.00a & 1.01a)
  sx40hs (1.00d, 1.00g & 1.00i)
  g1x (1.00e, 1.00f & 1.00g)
  g5x (1.00c, 1.01a, 1.01b)
  g7x2 (1.01a, 1.01b, 1.10b)
*

Offline reyalp

  • ******
  • 14128
Re: Real vs market ISO
« Reply #93 on: 28 / July / 2013, 20:07:39 »
Quote from: philmoz on 28 / July / 2013, 19:49:35
The Misc values for ISO are showing incorrect value for every CHDK ISO override, not just 80.

For example if I select Canon ISO 400 the Misc values shows I-R = 261, I-M = 400.
If I then use the CHDK override to select ISO 400 I see I-R = 241, I-M 369.

I would expect to see a visible difference in this case between the Canon and CHDK settings; but I don't.

This would seem to indicate that the Misc values display is broken.
It goes much deeper than that. If you look at the output I got from ISOTESTO / ISOOVER.LUA, the sv96 and sv96_market values that result from an override are lower than a correct conversion would give, across the board.

e.g. for D10, an ISO override of 80 gives sv96 368, sv96_market 437. That equates to a "real" ISO of 44.5 and a "market" iso of 73.3. Outside of rounding, this matches what the "misc values" show.

edit: and yes, this isn't related to the iso80 special case. It's a different problem. 100 ends up being 56 "real" 91 "market"

Like I keep saying, the existing override code is somewhat broken, but I don't want to touch that for 1.2. For 1.3, I want to have the *option* of getting it right, so I made a system that can accommodate the weird ISO 80 special case. If we decide we don't need it down the road, that's fine, we can strip out.

I did spend quite a bit of time trying to figure this stuff out...

edit: see this post http://chdk.setepontos.com/index.php?topic=10341.msg102856#msg102856 and attached spreadsheet.
« Last Edit: 28 / July / 2013, 20:11:08 by reyalp »
Logged
Don't forget what the H stands for.
*

Offline waterwingz

  • ******
  • 12140
    • Waterwingz
Re: Real vs market ISO
« Reply #94 on: 28 / July / 2013, 20:14:43 »
Quote from: reyalp on 28 / July / 2013, 20:07:39
I did spend quite a bit of time trying to figure this stuff out...

For my benefit,  please bear with me?  When I brought this topic up,  I was working on understanding / cleaning up the autoiso code.   In that code,  ISO values were entered in market.  In the ISO overrride and ISO bracketing they were entered in real.  If I follow this thread correctly,  both UI functions had issues that are now coming up,  and simply converting to everything market,  or everything real,  would not have solved that?
Logged
Ported :   A1200    SD940   G10    Powershot N    G16
*

Offline reyalp

  • ******
  • 14128
Re: Real vs market ISO
« Reply #95 on: 28 / July / 2013, 20:27:16 »
Quote from: waterwingz on 28 / July / 2013, 20:14:43
If I follow this thread correctly,  both UI functions had issues that are now coming up,  and simply converting to everything market,  or everything real,  would not have solved that?
The discrepancy  in the overrides doesn't have to do with the input values, it has to do with the sv96 and sv96_market values that are generated from a given input value.

Here's another way to look at it:

If you set ISO 100 in the canon UI and half press, the resulting values are
SV96 411
SV96_MARKET 480

If you convert those back to ISO using 3.125*(2(N/96)) you get a "real" iso of 60.7, and a "market" ISO of 100.

So this tells us:
1) The sv96 and sv96_market values that correspond to the cameras idea of ISO 100
2) 3.125*(2(N/96)) is the correct conversion to ISO. This holds for all the values in the menu.

Now if you enter 100 in CHDK ISO overrides, it's gets converted to 60.7 "real" but the resulting apex values are:
SV96 399 = ISO 55.7
SV96_MARKET 468 (this is actually market + delta) = ISO 91.7

As far as I know, it's always been this way. We just dismissed inconsistencies assuming it was the real<>market weirdness. Changing the input to "market" values doesn't affect this.

Again, this is not something I care about for 1.2. It's close enough.
Logged
Don't forget what the H stands for.
*

Offline philmoz

  • *****
  • 3450
    • Photos
Re: Real vs market ISO
« Reply #96 on: 28 / July / 2013, 21:36:25 »
Quote from: reyalp on 28 / July / 2013, 20:27:16
Quote from: waterwingz on 28 / July / 2013, 20:14:43
If I follow this thread correctly,  both UI functions had issues that are now coming up,  and simply converting to everything market,  or everything real,  would not have solved that?
The discrepancy  in the overrides doesn't have to do with the input values, it has to do with the sv96 and sv96_market values that are generated from a given input value.

Here's another way to look at it:

If you set ISO 100 in the canon UI and half press, the resulting values are
SV96 411
SV96_MARKET 480

If you convert those back to ISO using 3.125*(2(N/96)) you get a "real" iso of 60.7, and a "market" ISO of 100.

So this tells us:
1) The sv96 and sv96_market values that correspond to the cameras idea of ISO 100
2) 3.125*(2(N/96)) is the correct conversion to ISO. This holds for all the values in the menu.

Now if you enter 100 in CHDK ISO overrides, it's gets converted to 60.7 "real" but the resulting apex values are:
SV96 399 = ISO 55.7
SV96_MARKET 468 (this is actually market + delta) = ISO 91.7

As far as I know, it's always been this way. We just dismissed inconsistencies assuming it was the real<>market weirdness. Changing the input to "market" values doesn't affect this.

Again, this is not something I care about for 1.2. It's close enough.

So basically the functions 'shooting_get_sv96_from_iso' and 'shooting_get_svm96_from_iso' are broken.

Where did the equations for these come from?

Edit:
Actually it looks like 'shooting_get_svm96_from_iso' is correct, it is calculating:
    log2(ISO/3.125) * 96

But 'shooting_get_sv96_from_iso' is calculating:
    log2(ISO/3.364) * 96

Phil.
« Last Edit: 28 / July / 2013, 21:45:26 by philmoz »
Logged
CHDK ports:
  sx30is (1.00c, 1.00h, 1.00l, 1.00n & 1.00p)
  g12 (1.00c, 1.00e, 1.00f & 1.00g)
  sx130is (1.01d & 1.01f)
  ixus310hs (1.00a & 1.01a)
  sx40hs (1.00d, 1.00g & 1.00i)
  g1x (1.00e, 1.00f & 1.00g)
  g5x (1.00c, 1.01a, 1.01b)
  g7x2 (1.01a, 1.01b, 1.10b)
*

Offline reyalp

  • ******
  • 14128
Re: Real vs market ISO
« Reply #97 on: 28 / July / 2013, 22:07:12 »
Quote from: philmoz on 28 / July / 2013, 21:36:25
So basically the functions 'shooting_get_sv96_from_iso' and 'shooting_get_svm96_from_iso' are broken.
Yes that's what I've been saying. The other thing I've been saying is I don't want to muck with it for 1.2. It's wrong, but it's not hideously wrong, and it's no worse than what we were doing before.
Quote
Actually it looks like 'shooting_get_svm96_from_iso' is correct, it is calculating:
Then something else is messed up, because the resulting "market" ISO is still wrong (see my post above, 100 comes out as 91). Possibly the delta/base stuff.

Once we have that sorted out, *then* we can decide if the ISO 80 special case matters... I started down the road of making that correction before I understood how much else was wrong.

It's possible the shooting_get_sv96_from_iso is appropriate for one of the old cameras...
Logged
Don't forget what the H stands for.
*

Offline philmoz

  • *****
  • 3450
    • Photos
Re: Real vs market ISO
« Reply #98 on: 28 / July / 2013, 22:31:25 »
Quote from: reyalp on 28 / July / 2013, 22:07:12
Quote from: philmoz on 28 / July / 2013, 21:36:25
So basically the functions 'shooting_get_sv96_from_iso' and 'shooting_get_svm96_from_iso' are broken.
Yes that's what I've been saying. The other thing I've been saying is I don't want to muck with it for 1.2. It's wrong, but it's not hideously wrong, and it's no worse than what we were doing before.
Quote
Actually it looks like 'shooting_get_svm96_from_iso' is correct, it is calculating:
Then something else is messed up, because the resulting "market" ISO is still wrong (see my post above, 100 comes out as 91). Possibly the delta/base stuff.

Once we have that sorted out, *then* we can decide if the ISO 80 special case matters... I started down the road of making that correction before I understood how much else was wrong.

It's possible the shooting_get_sv96_from_iso is appropriate for one of the old cameras...

'shooting_get_sv96m_from_iso' is only used to get the 'base' market value - still trying to figure out what this is for.

If I change 'shooting_get_sv96_from_iso' to use the same equation as 'shooting_get_sv96m_from_iso' then I get the correct I-R & I-M the correct value for I-M displayed in Misc values; but now I-R is too high.

Not planning to change anything in 1.2; just trying to understand the mess.

Phil.
« Last Edit: 29 / July / 2013, 01:47:03 by philmoz »
Logged
CHDK ports:
  sx30is (1.00c, 1.00h, 1.00l, 1.00n & 1.00p)
  g12 (1.00c, 1.00e, 1.00f & 1.00g)
  sx130is (1.01d & 1.01f)
  ixus310hs (1.00a & 1.01a)
  sx40hs (1.00d, 1.00g & 1.00i)
  g1x (1.00e, 1.00f & 1.00g)
  g5x (1.00c, 1.01a, 1.01b)
  g7x2 (1.01a, 1.01b, 1.10b)
*

Offline waterwingz

  • ******
  • 12140
    • Waterwingz
Re: Real vs market ISO
« Reply #99 on: 28 / July / 2013, 22:46:56 »
Quote from: philmoz on 28 / July / 2013, 22:31:25
If I change 'shooting_get_sv96_from_iso' to use the same equation as 'shooting_get_sv96m_from_iso' then I get the correct I-R & I-M values displayed in Misc values.

Not planning to change anything in 1.2; just trying to understand the mess
Go for it.
Logged
Ported :   A1200    SD940   G10    Powershot N    G16
Pages: 1 ... 5 6 7 8 9 [10] 11 12   Go Up
« previous next »
 

Related Topics


SimplePortal © 2008-2014, SimplePortal