I propose some improvements to the vectorscope :
[1] Instead of a simple point, draw the four-squares little grid of strict confidence
[2] Draw also the four corners of large confidence aroud each point
[3] Think about adding the option to draw a bigger vectorscope, that occupy all the display space, in order to see it more precisely
[4] Think about adding the option to use a user-defined level of transparency for this tool
[5] Add the option for the marks to be valid for PAL or NTSC (user-defined switchable)
See here for an example : http://www.danalee.ca/ttt/images/7/vectorscope_bars_big.jpg
Many thanks !
the main problem is a vectorscope image that meets this requirements:
- clean line-art (we cant use the image you posted, its an image)
- license-free (we cant simply use a google'd image)
- correct resolution (to be defined - but 400px should be doable in fullscreen)
if you can draw a BMP image with magic lantern palette (see cropmark guides)
that matches your requirements, i can try to implement that.
Hi g3gg0 !
Good ! I will start immediately !
I will draw a small (like the existing one) and a bigger version, so it can be switchable by user willing.
Where can I send the works when done ?
Many thanks !
attach/link it to the forum.
please start with a 400px version and test it as a cropmark in LV mode.
if this version is displayed exactly how you wish it, it can be downscaled and saved to 200px or 100px variants.
hint: i think its a bad idea to use a lot of colors as many colors distract too much.
I'm following the cropmark tutorial you headed me to, so I'm working on a canvas of 720x480 resolution, so putting a big vectorscope on center for first version, and a smaller one on the upper left corner. Pls correct me if I'm doing it wrong, thankyou !
Also: it is stated that "color 3 is semi-transparent black", but it's not true: it is absolute covering black, like color 2 (starting numberig from zero, like in the cropmarks wiki help). I tried to load some ready-made cropmarks in the wiki, but they are faking transparency by dithering full-transparent pixels (color 0) with full covering pixels (color 2 or 3) in a chequered pattern. I'm stuck there...
I cannot find the color, in the cropmarks palette, that you used in your vectorscope as white-semi-transparent background: how and where I can find it ?
Update: I have done extensive tests and I modified the cropmarks wiki section: ERRATA: the first 80 colors are common, the others are camera dependent. CORRIGE: the first 20 colors are common. ERRATA: color 3 is semi-transparent. CORRIGE: only color 20 is semi-transparent.
Update: I can produce a full vectorscope 480x480 pixels with the ML palette and with transparent background (color 0), if it's ok for you, so you can put programmatically your white semi-transparent background: I can see I cannot put it by myself with the cropmarks palette, because the palette color number 20 is semi-transparent but more black-ish than yours.
480x480 pixels is real-full-screen (going under the upper and lower ML info stripes). If you want I can produce also a 400x400 pixels version (minus 40 pixels for upper stripe and minus 40 pixels for lower stripe, right ?). Then I can produce a 200x200 and a 100x100 version, so you can make a user-menu to choose the dimensions, that's ok ?
Work done !
Here is the 480px vectorscope : https://dl.dropbox.com/u/46130896/vector_480.bmp
Here is the 400px vectorscope : https://dl.dropbox.com/u/46130896/vector_400.bmp
Here is the 200px vectorscope : https://dl.dropbox.com/u/46130896/vector_200.bmp
Here is the 100px vectorscope : https://dl.dropbox.com/u/46130896/vector_100.bmp
It would be nice all the sizes can be user-choosen.
Also, because I used black for drawing the vectorscope, I think that you could use dark green to draw the vectors (like a real vectorscope).
I used full-transparent background, but if you can use your opaque-white background for the inside of the circle, it would be more readable.
Let me know !
Many thanks !
Some specifications:
[1] The BMPs are not RLE-compressed. If you need them to be RLE compressed, tell me so I will re-upload them.
[2] If you can head me to which palette color you used for your white-semi-transparent background used in the existent vectorscope (based on the original ML palette published on the cropmarks wiki page), I can semi-transparentize it by myself, so offering you the work already done.
... do someone can tell me if my works are ok, or they need to be revised ?
hi, i didnt implement it yet, give me 1-2 weeks.
Good ! Let me know when you implemented it, by replyng to this thread, thanks !
This is looks interesting. As somebody who wants a huge-ass vectorscope on the screen, what do you use it for exactly?
I understand what a vectorscope is, I'm just wondering in what situations people find them useful.
People use vectorscope to calibrate colors... bigger=more precise evaluation if they are out or not, and about what quantity. If you know what a vectorscope is, you should know. I use it to evaluate if colors are right and if they falls inside the correct ranges. In a thumbnailed vectorscope nobody can see this.
In what concrete situation would a vectorscope be useful though? And what do you use it for? Are you removing color casts? Correcting skin tones? Is it just about obtaining proper white balance? Or are you actually adding things to the frame to achieve some sort of balance of color in the photo?
http://en.wikipedia.org/wiki/Vectorscope
http://help.adobe.com/en_US/premierepro/cs/using/WSCE1EF46C-AB40-45f9-BAD8-D790421BD01A.html
http://www.videomaker.com/videonews/2012/04/13023-essential-video-skills-reading-a-vectorscope-and-waveform-monitor
http://www.kdenlive.org/users/granjow/introducing-color-scopes-vectorscope
http://www.danalee.ca/ttt/waveform_monitors_and_vectorscopes.htm
http://blog.ieba.com/wp-content/uploads/2010/06/ReadWaveformVector.pdf
http://www.youtube.com/watch?v=3VwDAtkdC74
I've added a 2x magnification option for the input signal, should be easier to see now, especially with flat picture styles.
The bitmaps and all those sizes are overkill IMO. The native size is 256 pixels, because that's the range of the U-V signal; doubling the entire thing will make it bigger than the screen. And making it smaller... it's already small enough.
At least implement one of the proposed little templates I have done, because in the Others you cannot see if the color is inside or outside the "legal" area, because simply there isn't such an area drawn in the original template, but only a single dot.
Also: if you switch to x2 magnification you have implemented, you no longer can see if the colors are really inside the safe areas or not, because you have to re-scale also the underlaying grid ! You can say: "who cares ?"...but this way a vectorscope is useless for the majority of applications.
I hope you agree ;)