Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion. Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules. |
2nd June 2015, 19:40
|
#41 | Link | |
|
Registered User
Join Date: Jan 2012
Location: Toulon France
Posts: 249
|
Thanks Gavino,
I have tried the yuv [16,235] <-> rgb [0,255] formulas, copied on Avisynth web page, but i did not know that 255 is not 255.0, like this short script show it. Thus i have not took this formula the last time. Quote:
|
|
|
|
|
2nd June 2015, 23:42
|
#42 | Link | |
|
HeartlessS Usurer
Join Date: Dec 2009
Location: Over the rainbow
Posts: 10,980
|
Quote:
__________________
I sometimes post sober. StainlessS@MediaFire ::: AND/OR ::: StainlessS@SendSpace "Some infinities are bigger than other infinities", but how many of them are infinitely bigger ??? |
|
|
|
|
|
11th June 2015, 19:02
|
#43 | Link | |
|
Registered User
Join Date: Jan 2012
Location: Toulon France
Posts: 249
|
Hello,
I have written this script to show my problem. I do not understand why RGBAdjust (working only with bias args) give a picture with look of coloryuv white balance, but whith differents luma an chroma values. I need advice. Thanks Quote:
|
|
|
|
|
|
11th June 2015, 21:06
|
#44 | Link |
|
HeartlessS Usurer
Join Date: Dec 2009
Location: Over the rainbow
Posts: 10,980
|
I imagine that it's mainly because of the colorspace conversions (16->235 -> 0->255 -> 16->235), ie
ConvertToRGB(matrix = "REC601") and ConvertToYV12(last, matrix = "Rec601") EDIT: Precision loss, pixel values are integer (also, Even though RGBAdjust takes float Bias args, cannot adjust pixels by fractional amount). I've added a line and commented out a couple (and changed order a little bit) Code:
ScriptClip( """
#------------ Color matrix input
Kr = 0.299
Kg = 0.587
Kb = 0.114
#------------ YUV values extraction
y = AverageLuma()
u = AverageChromaU()
v = AverageChromaV()
#----------- AWB corrections calcul
du = 128-u # in YUV range, correction is 128 subtract YUV channel value
dv = 128-v
#--------- Chroma values after automatic white balance
y_awb = y # ssS, Same as original
u_awb = u + du
v_awb = v + dv
#-------- RGB values for AWB chroma values
r_awb = (y-16)*255.0/219 + 2*(v_awb-128)*255.0/224*(1-Kr)
g_awb = (y-16)*255.0/219 - 2*(u_awb-128)*255.0/224*(1-Kb)*Kb/Kg - 2*(v_awb-128)*255.0/224*(1-Kr)*Kr/Kg
b_awb = (y-16)*255.0/219 + 2*(u_awb-128)*255.0/224*(1-Kb)
#------- For display target AWB Chroma values
# ssS, unnecessary colorspace conversion back to what it was before (above awb->rgb->awb)
#y_awb = (Kr*219/255.0)*r_awb + (Kg*219/255.0)*g_awb + (Kb*219/255.0)*b_awb + 16
#v_awb = 112.0/255.0*r_awb - Kg*112.0/255.0*g_awb/(1-Kr) - Kb*112.0/255.0*b_awb/(1-Kr) + 128.0
#u_awb = - Kr*112.0/255.0*r_awb/(1-Kb) - Kg*112.0/255.0*g_awb/(1-Kb) + 112.0/255.0*b_awb + 128.0
#------------ RGB values calculation
r = (y-16)*255.0/219 + 2*(v-128)*255.0/224*(1-Kr)
g = (y-16)*255.0/219 - 2*(u-128)*255.0/224*(1-Kb)*Kb/Kg - 2*(v-128)*255.0/224*(1-Kr)*Kr/Kg
b = (y-16)*255.0/219 + 2*(u-128)*255.0/224*(1-Kb)
#------ AWB RGB correction calculation
dr = r_awb - r
dg = g_awb - g
db = b_awb - b
#------ RGBAdjust bias use
ConvertToRGB(matrix = "REC601")
RGBAdjust(rb = dr, gb = dg, bb = db)
ConvertToYV12(last, matrix = "Rec601")
#------- For display output AWB Chroma values
y_out = AverageLuma(last)
u_out = AverageChromaU(last)
v_out = AverageChromaV(last)
Subtitle("target luma awb : " + string(y_awb)+ " target chroma u awb : " + string(u_awb)+ " target chroma v awb : " + string(v_awb)+\
"\nluma-output : " + string(y_out)+ " chroma u output : " + string(u_out)+ " chroma v output : " + string(v_out),lsp=20)
return last
""")
return last
and of course out of range YUV TV levels colors too, eg $000000. EDIT: Two page thread that is quite interesting: http://forum.doom9.org/showthread.php?t=92944
__________________
I sometimes post sober. StainlessS@MediaFire ::: AND/OR ::: StainlessS@SendSpace "Some infinities are bigger than other infinities", but how many of them are infinitely bigger ??? Last edited by StainlessS; 11th June 2015 at 23:10. |
|
|
|
|
12th June 2015, 18:24
|
#45 | Link | |
|
Registered User
Join Date: Jan 2012
Location: Toulon France
Posts: 249
|
Hi StainlessS,
Color space conversion can not explain the whole problem. I have modified my script to display output RGB channel values without ConvertToYV12. RGBAdjust use a channel standard deviation and thus apply a bias is not a simple value copy. RGB color space action is not easy. But RGBADjust and RGBAdapt work well. Quote:
|
|
|
|
|
|
13th June 2015, 06:42
|
#46 | Link | |
|
HeartlessS Usurer
Join Date: Dec 2009
Location: Over the rainbow
Posts: 10,980
|
Each pixel could be up to 0.5 in error for both of the colorspace conversions and also up to 0.5 in error for the RgbAdjust step.
What is the biggest discrepancy you are seeing ? I'm currently knocking up an RgbAdapt graffer thing, will post when done. EDIT: For RgbAdjust step, each and every pixel will suffer the same error (Bias float rounded to int before addition, for the each channel). Quote:
Code:
New_R = Min(Max(Old_R + Round(Bias_R),0),255)
__________________
I sometimes post sober. StainlessS@MediaFire ::: AND/OR ::: StainlessS@SendSpace "Some infinities are bigger than other infinities", but how many of them are infinitely bigger ??? Last edited by StainlessS; 13th June 2015 at 06:57. |
|
|
|
|
|
13th June 2015, 09:23
|
#47 | Link | |
|
Registered User
Join Date: Jan 2012
Location: Toulon France
Posts: 249
|
Hi StainlesS,
I have seen some differences for bias application, that i cannot explain with round error. I believe find the reason : channel standard deviation. But your answer i understand it is a mistake. I have written this short script to illustrate what i have seen. With a BlanckClip as video, channel standard deviation = 0, 100% of bias is in output. With any of my life videos, channel standard deviation > 0, only X % of bias is in input. I have only seen that with big bias. Thus in this example a bias value is 100. I do not know the rule, but it is good. With AWB bias computed in YUV color space and converted in RGB color space the RGB output picture look is very near the UV output picture look. Quote:
Last edited by Bernardd; 13th June 2015 at 09:28. |
|
|
|
|
|
13th June 2015, 21:56
|
#48 | Link |
|
HeartlessS Usurer
Join Date: Dec 2009
Location: Over the rainbow
Posts: 10,980
|
@Bernardd
Can I clarify earlier post, When setting Bias only and other args at default, then be equivalent to Code:
New_R = Min(Max(Old_R + Round(Bias_R),0),255) Code:
Private Sub Make_Lut(Lutix As Long)
' Lutix, 0 = Blue, 1 = Green, 2 = Red
Dim x As Long, i As Long, Bias As Double, Gain As Double, Cont As Double, RPow As Double, Spow As Double, v As Double, Lin As Long
Dim rGam As Double, sGam As Double, SPMid As Double
Dim pord As Boolean, Val As Long
Bias = Lut_Bias(Lutix)
Gain = Lut_Gain(Lutix)
Cont = Lut_Cont(Lutix)
RPow = Lut_RPow(Lutix)
Spow = Lut_SPow(Lutix)
pord = Lut_Pord(Lutix)
SPMid = Lut_SPMid(Lutix)
If RPow < 0.1 Then RPow = 0.1
rGam = 1# / RPow
If rGam < 0.004 Then rGam = 0.004
sGam = 1# / Spow
Bias = Bias / 256#
For i = 0 To 255
v = i / 255#
v = (v * Gain) + ((v - 0.5) * Cont + 0.5) - v + Bias
If (rGam <> 1# Or sGam <> 1#) Then
If (pord = False) Then
If (rGam <> 1#) Then ' Here, rpow 1st
If (v > 0#) Then ' avoid error
v = v ^ rGam
End If
End If
If (sGam <> 1#) Then
v = v - SPMid
If (v < 0#) Then ' bottom half
v = -v ' abs(), spin right way up
v = v / SPMid ' Make in range 0 -> 1.0
v = v ^ sGam ' spow
v = v * SPMid ' Back to original range
v = -v ' spin upside down again
Else
v = v / (1# - SPMid) ' Make in range 0 -> 1.0
v = v ^ sGam ' spow
v = v * (1# - SPMid) ' Back to range 0 -> 0.5
End If
v = v + SPMid
End If
Else
If (sGam <> 1#) Then ' Here Spow 1st
v = v - SPMid
If (v < 0#) Then ' bottom half
v = -v ' abs(), spin right way up
v = v / SPMid ' Make in range 0 -> 1.0
v = v ^ sGam ' spow
v = v * SPMid ' Back to range 0 -> 0.5
v = -v ' spin upside down again
Else
v = v / (1# - SPMid) ' Make in range 0 -> 1.0
v = v ^ sGam ' spow
v = v * (1# - SPMid) ' Back to range 0 -> 0.5
End If
v = v + SPMid
End If
If (rGam <> 1#) Then
If (v > 0#) Then ' avoid error
v = v ^ rGam ' rpow
End If
End If
End If
' Avoid possible overflow due to multiple powers
If (v > 1#) Then
v = 1#
ElseIf v < 0# Then
v = 0#
End If
End If
v = (v * 255#) + 0.5
Val = Int(v) ' Round towards -ve infinity
If (Val > 255) Then
Val = 255
ElseIf Val < 0 Then
Val = 0
End If
Lut(Lutix, i) = Val
Next i
End Sub
Where 'off' is equivalent to Bias. Code:
Private Sub Make_Lut(Lutix As Long)
Dim i As Long
Dim off As Double, gain As Double, gamma As Double
Dim v As Double
off = Lut_off(Lutix)
gain = Lut_gain(Lutix)
gamma = 1# / Lut_gamma(Lutix)
If gamma < 0.004 Then gamma = 0.004
For i = 0 To 255
v = (off + i * gain) / 255#
v = IIf(v < 0#, 0#, IIf(v > 1#, 1#, v))
Lut(Lutix, i) = Int((v ^ gamma) * 255# + 0.5)
Next i
End Sub
RgbAdapt() zip updated 13 june 2015, Added RgbAdapt_Graffer executable to demo plugin. Dll not updated from v0.1. See 1st post. Fish:  Frog: Showing Line=2, ie bubbles rather than straight line graph [EDIT: Shows big jump for both red and green in LUT]. EDIT: I should have moused over the graphic for both jpg's to show Before and After RGB colors under the mouse. Zip includes Graffer source in VB6. Fish2b showing Full Grid and Before & After Colors.  EDIT: The 'Copy To Clipboard' button copies the RgbAdapt() function call together with arguments. The Left half of frame is corrected (After and Right Half Before). You can see my Windows 95 RedTile's,  EDIT: And with full frame modified [Half Screen unticked]: 
__________________
I sometimes post sober. StainlessS@MediaFire ::: AND/OR ::: StainlessS@SendSpace "Some infinities are bigger than other infinities", but how many of them are infinitely bigger ??? Last edited by StainlessS; 15th July 2017 at 18:55. |
|
|
|
|
15th June 2015, 21:44
|
#49 | Link | |
|
Registered User
Join Date: Jan 2012
Location: Toulon France
Posts: 249
|
Hi StainlessS
I have read RGBAdapt source in the past, but it is difficult for me to understand formulas in vb6. Thus now with "lut" i can not say if it is a variable or a instruction. If you have internet page to learn,thank you. Your graffer is a nice tool. Perhaps, it will be a few better with display of RGB values before and after. But I have used it to analyse my manual process. I make iteration. Thus i propose this new version, which add linear correction to gamma correction. Linear correction is based on the output of gamma correction. Quote:
|
|
|
|
|
|
16th June 2015, 11:15
|
#50 | Link |
|
HeartlessS Usurer
Join Date: Dec 2009
Location: Over the rainbow
Posts: 10,980
|
A LUT (Look Up Table) is just an array that is pre-prepared, where you LATER get the modified color using the
original channel value as an index into the table. EDIT: Lutix is just the RGB channel number, 0=B, 1=G, R=2. It is quicker to set up a 3*256 pre-calculated array than to do the calculations on each and every pixel for each channel. I screwed up the display of 'Before' and 'After' colors, I forgot that RGB is upside down when I added that. Added two little boxes show the actual color as well as Hex values. Have added 4 more arguments for each channel, IMin, OMin, IMax, and OMax (all 0->255, IMin <= IMax). Any INPUT value less than IMin will be mapped to OMin, and values greater than IMax will be mapped to OMax. Only input values that do not comply with above are processed as usual. This allows for weird shaped LUT to eg create masks. Made a few more changes to the Graffer, will update a bit later today (I think) EDIT: Both Graffer and dll pending update. EDIT: Previously posted RgbAdjust Graffer in VB 6 converted from Avisynth CPP source, with comments Where 'off' is equivalent to Bias (in RgbAdapt). [EDIT: Actually its called eg rb rather than off or R_Bias in RgbAdjust, Off is just what I called it in the RgbAdjust graffer code]. Code:
Private Sub Make_Lut(Lutix As Long)
Dim i As Long
Dim off As Double, gain As Double, gamma As Double
Dim v As Double
off = Lut_off(Lutix) # Below just getting Off(bias) Gain and Gamma args
gain = Lut_gain(Lutix)
gamma = 1# / Lut_gamma(Lutix)
If gamma < 0.004 Then gamma = 0.004
For i = 0 To 255 # Process 0 to 255 in the LUT array
v = (off + i * gain) / 255# # apply Gain and Bias and range 0.0 -> 1.0
v = IIf(v < 0#, 0#, IIf(v > 1#, 1#, v)) # This is just a Min(Max()) 0.0 -> 1.0 operation, 1# just means a Double Float ie same as 1.0
Lut(Lutix, i) = Int((v ^ gamma) * 255# + 0.5) # Apply Gamma, range 0 -> 255, store in LUT array for Lutix chan(2 dim array, 3x256)
Next i
End Sub
Code:
NEW_R = RED_LUT(OLD_R) # where OLD_R is original Red channel value Code:
New_R = Min(Max(Round(Old_R + Bias_R),0),255) Code:
New_R = Min(Max(Old_R + Round(Bias_R),0),255)
__________________
I sometimes post sober. StainlessS@MediaFire ::: AND/OR ::: StainlessS@SendSpace "Some infinities are bigger than other infinities", but how many of them are infinitely bigger ??? Last edited by StainlessS; 17th June 2015 at 03:49. |
|
|
|
|
16th June 2015, 18:43
|
#51 | Link | |
|
Registered User
Join Date: Jan 2012
Location: Toulon France
Posts: 249
|
Hi StainlessS,
Thank for explanation. This day, i have used the graffer, and i have found a error conception for my script. I have linked Rpow and Spow corrections. For little correction is not a real problem, but for big is not good. I believe that i have smelt this problem, because i have put in the old script a pseudo gamma variable, which can be increase if neccessary. With your graffer, i can analyze the process with a external view. For gamma curve the RGB difference must be converted by formula pow(4.0, RGB difference) to give the gamma correction value. With this formula, i can have similar power action in automated white balance and graffer white balance with bias and rpow args tunes. But with this formula Rpow and Spow link is no good for little corrections. Thus i have written a new version where the spow args are only manual. It is like a automated RGBAdjust with manual inflexion curves if necessary. For normal video automated correction have a good look. For bad video add manual correction can be nice. This day, now, i have not idea to automate Spow action. Quote:
|
|
|
|
|
|
17th June 2015, 03:23
|
#53 | Link |
|
HeartlessS Usurer
Join Date: Dec 2009
Location: Over the rainbow
Posts: 10,980
|
Cant do update just yet.
For some reason my VS6 compiler is producing code that results in access violations, only for release mode (not DEBUG) and not when using VS ToolKit 2003. Narrowed it down to about 10 lines of code the only effect of which was to link in an avisynth dll, and when called produce a ThrowError() error message, ie "env->ThrowError("TEST: This is a test error");". Was managing to produce access violation before getting there. [EDIT: But only in some media players eg WMP, not Vdub]. Finally tried a compile on another machine and no error exception produced by resultant dll. (same VS6 compiler). [EDIT: And tested on same machine that was producing original error.] When I started using VB6 a few days back, it demanded that I give it the Platform SDK Disk to update, and I think it somehow screwed my CPP compiler. Anyways, looks like I might be doing a system recovery from stable image from a few months back, I post update as soon as I can. If you would like to try out the (much improved I think) Graffer, here tis:- EDIT LINK REMOVED EDIT: kuchikirukia, DL the graffer and load in the image provided, you will easily see if the dll can be of use (should not be too long till I update, perhaps tomorrow as I could [and usually do] do final compile using VS ToolKit 2003 which is not affected by the problem I'm having. I have to get some sleep now as its 03:30AM). EDIT: The image YOU provided.
__________________
I sometimes post sober. StainlessS@MediaFire ::: AND/OR ::: StainlessS@SendSpace "Some infinities are bigger than other infinities", but how many of them are infinitely bigger ??? Last edited by StainlessS; 17th June 2015 at 19:49. |
|
|
|
|
17th June 2015, 04:19
|
#54 | Link |
|
HeartlessS Usurer
Join Date: Dec 2009
Location: Over the rainbow
Posts: 10,980
|
k, here spent about two minutes playing with the sliders, you could do better.
Maybe start with about same settings Ive used and also play with SPMid settings. I've used identical for R,G and B. Before on the right, mod on the left. You do not have a lot to play with. Here:-  Gonna get some sleep now. EDIT: The currently available dll has all args execept the bottom 4 (which are unlikely to be of use in this case). You will want to reduce the lighter levels a bit from what I've used. EDIT: If you do get a setting that you are happy with, post result here, I'de be interested to see what you can get out of the sample. EDIT: I'll try to implement a Lock, so you can simultaneously set all channels the same, its real tricky one at a time. EDIT: is it a single frame or video ? Is camera stationary ? Maybe need answer to both (well 2nd i video), so others could assist. Perhaps needs some kind of mask if stationary or single frame, lighten audience and you screw with the stage lowlites.
__________________
I sometimes post sober. StainlessS@MediaFire ::: AND/OR ::: StainlessS@SendSpace "Some infinities are bigger than other infinities", but how many of them are infinitely bigger ??? Last edited by StainlessS; 15th July 2017 at 18:56. |
|
|
|
|
17th June 2015, 19:52
|
#55 | Link |
|
HeartlessS Usurer
Join Date: Dec 2009
Location: Over the rainbow
Posts: 10,980
|
RgbAdapt v0.3, dll and Graffer updates., see 1st post.
Added IMin, OMin, IMax, OMax args for each channel. Now copies rather than zero Alpha channel (v0.2 no public release). Code:
RGBAdapt v0.3, By StainlessS
For Avisynth v2.58+
RGB24/RGB32.
Audio as original.
RGBAdapt(clip c, \
\ Float "R_Bias"=0.0,Float "R_Gain"=1.0,Float "R_Cont"=1.0,Float "R_RPow"=1.0,Float "R_Spow"=1.0,Float "R_SPMid"=0.5,Bool "R_Pord"=false,
\ Float "G_Bias"=0.0,Float "G_Gain"=1.0,Float "G_Cont"=1.0,Float "G_RPow"=1.0,Float "G_Spow"=1.0,Float "G_SPMid"=0.5,Bool "G_Pord"=false,
\ Float "B_Bias"=0.0,Float "B_Gain"=1.0,Float "B_Cont"=1.0,Float "B_RPow"=1.0,Float "B_Spow"=1.0,Float "B_SPMid"=0.5,Bool "B_Pord"=false,
\ Int "R_IMin"=0,Int "R_OMin"=0,Int "R_IMax"=255,Int "R_OMax"=255,
\ Int "G_IMin"=0,Int "G_OMin"=0,Int "G_IMax"=255,Int "G_OMax"=255,
\ Int "B_IMin"=0,Int "B_OMin"=0,Int "B_IMax"=255,Int "B_OMax"=255
\ )
Bias -512.0 -> 512.0. Bias = Brightness.
Gain -8.0 -> 8.0 Gain = Zero relative contrast.
Cont -8.0 -> 8.0 Cont = Centre relative Contrast.
RPow 0.1 -> 4.0 RPow = Gamma, r shaped power curve.
SPow 0.1 -> 4.0 SPow = S shaped power curve.
SPMid 0.01 -> 0.99 SPMid = Controls Inflection point for S shaped power curve, 0.5 = mid point ie 0.5*255.0
Pord True/False Pord=Rpow processed first (False) or SPow first (True). The power functions are non commutative.
Assuming all settings apart from RPow, SPow and SPMid at defaults:-
Pord Default (false) RPow1st, when modifying RPow or SPow, the SPow power curve inflection point will
remain fixed at Y = SPMid * 255.0, Rpow will slide it left or right.
Pord SPow1st, when modifying RPow or SPow, the SPow power curve inflection point will remain fixed
at X = SPMid * 255.0, RPow will slide it up or down.
New args v0.3
IMin 0 -> 255 : IMin <= (IMax+1)
OMin 0 -> 255
IMax 0 -> 255 : (IMin-1) <= IMax
OMax 0 -> 255
Any INPUT value less than IMin will be mapped to OMin (both default to 0), and values greater than IMax will be mapped to
OMax (both default to 255). With default settings the Min/Max options have no effect.
These options allow for weird shaped LUT's to eg create masks. Example, Leaving all other settings at default and
for some channel set IMax to eg 127 and OMax to 64 would map all input above 127 to 64 while leaving everything else untouched.
If you additionally set eg IMin=128 and OMin=180, then 0-127 would map to 180 and 128->255 map to 64.
The IMin <= (IMax+1) requirement allows you to map the entire 0->255 range to any two output values.
RgbAdapt_Graffer.exe demos the plugin and all arguments. (Will require VB6 runtimes, most will already have these.)
__________________
I sometimes post sober. StainlessS@MediaFire ::: AND/OR ::: StainlessS@SendSpace "Some infinities are bigger than other infinities", but how many of them are infinitely bigger ??? Last edited by StainlessS; 17th June 2015 at 21:54. |
|
|
|
|
17th June 2015, 22:36
|
#56 | Link |
|
HeartlessS Usurer
Join Date: Dec 2009
Location: Over the rainbow
Posts: 10,980
|
Here a better daft example what the Min/Max args could be used for (mask making) [frog not really part of example].
(not intended to do anything useful, just to show what you can do)  Also NOTE, RgbAdapt Graffer can be used with RgbAdjust (Bias[integer only], Gain and RPow only). EDIT:Actually, Bias implemented already as Float in both dll and Graffer.
__________________
I sometimes post sober. StainlessS@MediaFire ::: AND/OR ::: StainlessS@SendSpace "Some infinities are bigger than other infinities", but how many of them are infinitely bigger ??? Last edited by StainlessS; 16th March 2019 at 11:42. |
|
|
|
|
17th June 2015, 23:03
|
#57 | Link |
|
Registered User
Join Date: Oct 2014
Posts: 476
|
I'm thinking something like this, maybe.
 Preview is too small to really tell and I have another encode going so I'm not ready to get on this in avisynth. Source is actually a commercial Blu-ray. I think they killed the levels because they projected onto a glossy screen which gave a hilariously enlarged and distorted reflection of the audience. So Miku is performing in a sea of demonic red faces. |
|
|
|
|
17th June 2015, 23:16
|
#58 | Link | ||
|
HeartlessS Usurer
Join Date: Dec 2009
Location: Over the rainbow
Posts: 10,980
|
Quote:
(which makes it actually a bit less smooth looking than if 256x256 [somewhat like point resize, sort of]). EDIT: With 512x512 get better view when line in Dots/Bubbles style, to see jumps/gaps in LUT. With 512x512 the graf consists of lots of little line segments without some overall formula (with rounding) to achieve required curve. With 256x256, for the most part the lines can be up/down one pixel, left or right 1 pixel or 1 pixel 45 degree diagonal, looks quite smooth curve because of this. Quote:
I was trying to bring out more of the audience, silly me EDIT: It's not too difficult to copy the RgbAdapt function call to clipboard (via the Button) and CTRL/V paste into eg VDubMod script editor and press VD refresh button to view in all its glory.
__________________
I sometimes post sober. StainlessS@MediaFire ::: AND/OR ::: StainlessS@SendSpace "Some infinities are bigger than other infinities", but how many of them are infinitely bigger ??? Last edited by StainlessS; 18th June 2015 at 01:16. |
||
|
|
|
|
18th June 2015, 11:15
|
#59 | Link | ||
|
HeartlessS Usurer
Join Date: Dec 2009
Location: Over the rainbow
Posts: 10,980
|
Quote:
(dont know why, thought I'de done it as int in both dll and graffer). EDIT: @ Bernardd I see that you've left out all of the Float(Default()) floats again  Quote:
 http://forum.doom9.org/showthread.php?t=154731
__________________
I sometimes post sober. StainlessS@MediaFire ::: AND/OR ::: StainlessS@SendSpace "Some infinities are bigger than other infinities", but how many of them are infinitely bigger ??? Last edited by StainlessS; 18th June 2015 at 12:56. |
||
|
|
|
|
18th June 2015, 23:36
|
#60 | Link | ||
|
Registered User
Join Date: Jan 2012
Location: Toulon France
Posts: 249
|
Hi StainlessS
Below, the last version. Default, white balance is automatic with Rpow curve. If it is not enought, first you can enable Spow curve (spmid and pord are auto), secund you can enable auto gain and cont args. You have perhaps to choice the processed color channels by gain and cont. You can always mix classic awb (linear) with gamma awb (the only float(default)  ).You can also display the computed values to use as start values for RGBadapt manual use. Quote:
This work is in RGB color space for compute in RGB color space, do you think it is an error ? I need time to digest these datas http://forum.doom9.org/showthread.php?t=154731 Quote:
Last edited by Bernardd; 18th June 2015 at 23:42. |
||
|
|
|
 |
| Tags |
| rgbadjust |
|
|
Linear Mode
