Joined: 11/18/2006
Posts: 2426
Location: Back where I belong
I still see mostly 0's for tech score for the only couple of movies to have been published since the new rating system was implemented:
http://tasvideos.org/rating.exe/1196/detailshttp://tasvideos.org/rating.exe/1195/details
Any particular reason why this are still goofy? Is it because those particular tech ratings have decimals, and the published movie rating system doesn't accommodate decimals yet?
I still see mostly 0's for tech score for the only couple of movies to have been published since the new rating system was implemented:
http://tasvideos.org/rating.exe/1196/detailshttp://tasvideos.org/rating.exe/1195/details
Any particular reason why this are still goofy? Is it because those particular tech ratings have decimals, and the published movie rating system doesn't accommodate decimals yet?
After examining the database, I think I do:
For example IronSlayer's votes on movie 1196:
mysql> select * from movie_rating where movieid=1196 and userid=1282;
+--------+---------+---------------+-------+
| userid | movieid | ratingname | value |
+--------+---------+---------------+-------+
| 1282 | 1196 | Entertainment | 5.00 |
| 1282 | 1196 | Tech Quality | 6.00 |
+--------+---------+---------------+-------+
However, mmbossman's votes on the same movie:
mysql> select * from movie_rating where movieid=1196 and userid=1018;
+--------+---------+---------------+-------+
| userid | movieid | ratingname | value |
+--------+---------+---------------+-------+
| 1018 | 1196 | Entertainment | 5.00 |
+--------+---------+---------------+-------+
No technical score is being stored into the database for some reason. I assume that the PHP code is thus getting zeros for them.
mmbossman: Is this problem happening when submitting from the ratings page, or are these the votes from the movie submissions forum?
Joined: 11/18/2006
Posts: 2426
Location: Back where I belong
I have always rated both entertainment and technical categories when I vote, so I went back to check the submission thread. There's only 1 vote, with a 5 for entertainment, and a 7 for technical, and I believe that is what I rated it. So this seems to be a problem with the ratings from the submission not being transferred over once the movie is published.
I have also never had a problem with rating a movie after it has been published, but unless 4 other people actually rated that movie as 0 in technical, it can't be a problem solely from transference of ratings from the submission thread (since there was only 1 rating, which happened to by mine).
I just submitted a rating for that movie, and both ratings were properly stored in the database.
I can't conclude anything else than that the problem is with the movie submissions forum voting. For some reason that code is not always storing the technical score in the database (although it *does* store it most of the time, as seen from the recent SuperMario64 movie).
I would have guessed that if someone does not vote on one of the categories in the submissions forum, then the script doesn't store anything for that category. However, if you did indeed vote on both categories in the submission of this movie in question, then something odder must be going on.
Given that the submission forum voting code was not written by me I think Bisqwit should take a look at it.
Edit: But anyways, it should indeed be decided what to do if someone does not vote on one of the categories. I don't know how Bisqwit intended this to work, but I suppose that if the vote for one of the categories does not appear in the database, a "-" (rather than a zero) could be printed wherever this is used. Of course this is used in several places in the code. I wonder if it could be somehow more automatized then writing explicit conditionals at each place. (Also the calculating the averages should take this into account, but I don't know if Bisqwit has already done that.)
Joined: 4/20/2005
Posts: 2161
Location: Norrköping, Sweden
I'm not sure this is the topic to report in, but I found something strange with the number of votes for the newly published SM64 run, that might have something to do with this new voting system:
How can a run have 78.5 votes? Is the 0.5 vote for a vote that rated only entertainment or only technical? Again, sorry if this is the wrong thread.
How can a run have 78.5 votes? Is the 0.5 vote for a vote that rated only entertainment or only technical? Again, sorry if this is the wrong thread.
AFAIK the script counts all the ratings given to the movie (entertainment and tech quality are two distinct ratings) and divides it by the amount of rating types, ie. 2. For this reason if someone only votes for one of the categories, it will only count as a half vote.
(Bisqwit: Looking at the code in question, I think it might have a bug: If everyone voted eg. only on Entertainment and not on Tech Quality, it would not divide the number of votes by 2, but by 1 instead. If even one person then votes for Tech Quality, it will then start dividing by 2.)
Ok, now if there's no rating for one category in the database, a "-" will be shown rather than a zero. Also the technical ratings should again be visible in the movie statistics page.
I still don't know if there's a bug in the submission of ratings somewhere. Please report if you are sure you have submitted a rating in the submissions forum but a "-" appears as your rating in the ratings list for that movie (when the movie is published).
That's very strange. I precisely went through that code in order to make it work with decimals, and I tested it. But it indeed is not showing the decimal. I can't understand why. I can't see any problem in the code... :/
I wonder if it might be some problem with caching (if I'm not mistaken the server caches pages rather than generating them every time they are requested). Although I have never encountered such a problem before...
We'll have to see if the problem persists and is repeatable...
Edit: It seems that it was just a problem with versioning. The new code should be in place again soon.
Or we could teach everyone the awesome english word "Abstain"
It doesn't seem to work very well:
http://tasvideos.org/forum/viewtopic.php?t=7413http://tasvideos.org/forum/viewtopic.php?t=7416
EDIT: I guess people who don't know this word may assume that dragging the bar straight to 0 can do the trick.
EDIT 2: It can be that some people just hate the movies, but an extra "No Opinion" button to cancel out one's own votes may help things.
EDIT 3: It could be caused by an old bug:
The voter can be deceived with the checked "Abstained", but acctually the vote is 0 for Tech.
It has already been fixed by Bisqwit, though.
<klmz> it reminds me of that people used to keep quoting adelikat's IRC statements in the old good days
<adelikat> no doubt
<adelikat> klmz, they still do
For your information, I fixed a bug with the rating indicator that caused low ratings (indicated in blue tones) to be indicated in too bright a colour, often causing a bright white spot on the low region even when relatively few ratings actually occur on that region.
It was a miscalculation in the mixture of HSV & YUV colorspaces aiming to calculate a realistic brightness value representing the rating's strength in that particular region.
Here's how the colours work.
On X axis, is the hue. It indicates the value of the rating. 0 = on the left side, 10 = on the right side.
On Y axis, is the brightness. It indicates the popularity of that particular rating option. Top = Nobody has voted that option, bottom = most people have voted that option.
Here's how the colours used to be indicated, when it was buggy.
Here's how they are indicated after the bug was fixed.
Here's how they would be indicated if the relative brightness differences of red, green and blue were not observed in the calculation (i.e. a full blue would be treated as equally bright as a full green):
I'm kind of torn how to actually go ahead with the colours :I
I'm kind of torn how to actually go ahead with the colours :I
Whilst different colours do have different brightnesses, the 'after it was fixed' example looks far more uneven to me - as if you did too much. The bottom example seems almost right - any tweaking needed is minute.
I'm just some random guy. Don't let my words get you riled - I have my opinions but they're only mine.
I think that the second image looks "uneven" only because of the way it's laid out and how the three color components are weighted according to their brightness.
The idea is that any given horizontal line of pixels in that image has the same perceived brightness. (Of course getting a perfect result for this is impossible, as it depends on a million things, including the physical properties of your display, the gamma correction your software is applying to the image, the gamma correction your hardware is applying to the image, your monitor settings and even the slight differences in perception of color and brightness the human eye has from individual to individual.)
When all the horizontal lines are grouped together like that, you will immediately see the diagonal patterns emerging from the color channel weighting. However, these patterns do not appear if you take one single horizontal line of pixels from that image. Instead, that line of pixels should have an (almost) even brightness throughout. Which is the idea.
The third image may look more "pleasing" as laid out, but if you examine it carefully you will notice that the brightness of individual horizontal pixel lines varies greatly (in this case being dimmest on the sides and brightest on the cyan and yellow spots).
I think the second image is the most correct for the intended usage.
I ran the images though Irfanview's "Convert to grayscale" option. The results are <cut>
Wow!
That's amazing! Thank you so much for posting this!
I guess that settles it.
P.S. Warp, you're totally right. I just spent a few minutes using two bundles of folded paper and moving them up and down so I could only see a thin horizontal strip between the 2.
Image 2 IS totally perfect and image 3's issues become far more apparent when seen this way. This is the most astonishing thing I've seen all month! Thank you both!
I'm just some random guy. Don't let my words get you riled - I have my opinions but they're only mine.
This actually makes sense only if your monitor has just the right gamma :)
On my laptop's LCD, if you watch it from an upwards angle, it starts looking right, but if I watch it straight, not so.
This actually makes sense only if your monitor has just the right gamma :)
If it doesn't, you only have the monitor to blame and image 3 is likely even worse.
Using the 'view-it-between-two rectangular-objects-method' I now see how much better #2 was than#3.
I'm just some random guy. Don't let my words get you riled - I have my opinions but they're only mine.
How can a run have 78.5 votes? Is the 0.5 vote for a vote that rated only entertainment or only technical? Again, sorry if this is the wrong thread.
The voter can be deceived with the checked "Abstained", but acctually the vote is 0 for Tech.
It has already been fixed by Bisqwit, though.
Here's how they are indicated after the bug was fixed.
Here's how they would be indicated if the relative brightness differences of red, green and blue were not observed in the calculation (i.e. a full blue would be treated as equally bright as a full green):
I'm kind of torn how to actually go ahead with the colours :I
