Saturday, August 30, 2008

// // Leave a Comment

Canon digital still-camera captures motion better than the camcorders!

Abstract:

Wanna make indipendent films without jagged motion?
Wanna see videos in full resolution on a computer screen?

Get a professional camcorder or... Canon digital still-camera!!!

A paradox, but seems to be true for the most camcorders... :)

See frames-comparison at the end of the article.


I've been in a great hesitation what digital camcorder to buy in order to make my movies with, but I still stick to my good-old Canon Powershot A540 still-camera. And it is for a reason...

What digital camcorder to buy - a buying guide by Tosh

The main concerns were video quality and ease of frame-by-frame editing, camera is intended mostly for movie-making. ( Well, it wasn't so simple, but let's simplify it. :) )

So what kind of storage to use?

1. Mini-DV
2. DVD
3. HDD
4. SD Card


It seemed, that my main goals directly lead to MiniDV with its inter-frame compression of 25 Mbit/s and frame-by-frame!

However, mini-DV video is interlaced and the motion looks terrible on a PC screen - my movies wouldn't make it to TV directly :)

So video should be de-interlaced. But after being deinterlaced, the video looses one half of its vertical resolution and it doesn't look nice. Thus with miniDV you seem to get higher quality, which however should be deinterlaced and screwed up???

What about the other formats?

DVD, SD, HDD - usually use MPEG2, up to 9 Mbit/s. Must be decoded during editing, which is less convenient, decompression and recompression leads to more quality losses, and fast motion is supposed to look bad.

However, I thought that MPEG2 cameras are deinterlaced on-the-fly, so this might be better?
I thought so, until I did play with a 2007 Sony HDD camera - Sony DCR SR-200.

Interlaced, also!!! Terrible... But is it like that for all?

I am afraid that in order to be satisfied with the look of the motion of the video, I need either progressive scan, i.e. professional camcorder or Canon HD pro-sumer camcorder. In the worst case, I have to stick to my wonderful piece of engineering Canon A540 or get a newer model. :)

Canon Powershot A540 (2006) - Professional video with digital still-camera!

- MJPEG, about 15 Mbit/s (tweakable by CHDK), allowing painless frame by frame editing.
- Progressive scan and nice looking motion blur.
- Convenient reviewing of videos and even editing with the camera itself.

Disadvantages:

- No image stabilization.
- Limited zoom - only 4x and no good control of its speed (vs. like 10x - 45x and control of the speed of zoom)
- No exposure control and manual focus.
- Just 11 kHz mono sound
- Limited storage - maximum 2 GB SD cards, less than 20 minutes per card.
- Slightly lower resolution (640 x 480 vs 720x576)


...

Video with Digital Still-Camera (Canon A540)

Note the smooth motion-blur








Video with a camcorder (Sony DCR SR-200, HDD)




PS. Actually, while playing those "ghosting" effect is not so noticeable... But if you want to take a frame from the video - surpriiiiiize!

Whatever... Interlacing sucks... ;-)
Read More

Friday, August 29, 2008

// // Leave a Comment

A Start-up or a PhD? - that is the question

That is the question I started to ask myself like an year ago, when I was about to finish my BS.
I've always wanted to create my own AI-research company where to develop my theories... Always means at least since my late teenage years. However, this is still a dream. :)

I've met several appropriate people during the years, but it was just an exchange of thoughts....

Last year I realized something which sounded a bit like a paradox: what is better - to create your own research-company after you are a BS or an MS, or to go for a PhD. My first preference was the start-up...

If you manage to reach to a fundamental or feasible discovery/theory and have a basis for an impressive PhD, then you may have the ability to create a company and directly realize your ideas into a product which can go to the market, bring you new funding and allow you to do more research.

If you have your ideas and topics and you want to do a PhD,  you  first need to find a University with the appropriate research direction, which might happen to be hard or impossible... I don't think it's a good idea to do any PhD, I would select it with attention.

If you find a right place, it is possible that what you develop during your studentship would not be your-property, and if you develop something really promising, it might be a disadvantage, you may expose your  ideas to more powerful authorities that can steal your idea and create the product before yourself.

Finally, while doing a PhD you are supposed to demonstrate that you can perform significant research and contribution by yourself, etc., while if you are in a company, you can freely collaborate and potentially scale your R&D team.

In the same time, finding of funding in the early stages of projects is a hard task and it is likely that during a PhD you will lay the milestones of the future development...

A Start-up or a PhD?

What about both? Create a company and do a PhD in parallel?

Capital...

Of course, if you don't have funding, you can't found a start-up... You need to do something for a living and you're eating up the time for the real R&D.

So doing a PhD, even not the "dream one" may help you to concentrate on research.

Success or Failure PhD paradox

If you have an ingenious project, it is likely to find funding even without a PhD.
If you have a project and a basis which are not impressive enough and are yet only of Academic interest or it sounds impressive only to scientitsts, then you're supposed to do a PhD...

So if you sort of fail, you go to a PhD to get a higher formal degree and eventually direct to Academia, or in the meantime you reach to something better.

If you succeeded early, then a PhD is not required for your future success, your capability to do "self-guided significant blah-blah-blah" is obvious.

Sure... :)

....

Update, 5.9.2022: See a continuation of this story:

PhD system and duration: 5 years with 3 years planning - isn't it too long and risking obsolescence on the go? What about contributing to many projects and not just one "personal" thesis?   https://artificial-mind.blogspot.com/2022/09/phd-planning-3-or-5-years-phd-format.html


Read More

Thursday, August 21, 2008

// // 1 comment

The Moskvitch Driver (The Looser) | Шофьорът на москвич (Нещастникът)



Пробна сцена за един или за няколко от филмите/клиповете, по които работя в момента и не съм им уточнил заглавията...

A test sequence for one or several short films/music videoclips I'm working in progress. Not dubbed, yet.

It was real fun, after me and Maria "maiova" Ivanova shot the photo story "My Love"... http://artificial-mind.blogspot.com/2008/07/my-love.html

Hope you'll enjoy it too! :)

Режисьор, Монтаж, Актьор: Тодор Арнаудов
Оператори: Мария Иванова и Тодор Арнаудов

Director, Editor, Actor: Todor Arnaudov
Camera: Maria Ivanova and Todor Arnaudov

Equipment: Canon Powershot A540 :)




High Quality 640x480: http://toshuniverse.com/video/test/cialo-moskvitch-dulgo-1-c3.avi

Youtube: http://www.youtube.com/watch?v=mqELYA-HUvM


Yep, it's shaking in the beginning, but unfortunately it's almost impossible to reshoot that sequence.
Read More

Saturday, August 16, 2008

// // Leave a Comment

White bread is killing neurons?! | Белият хляб убива нервни клекти?!

English - see below

Много се смях, когато видях този надпис в Англия:


Приятелят ми, с когото се разхождахме в парка обаче отбеляза, че белият хляб бил много вреден и за хората, ама... Не се замислих с какво точно е толкова вреден. Че е по-мазен? Или че по-трудно се храносмила?



После имах колеги, които не ядат хляб, и ми обясняваха за вредата от прясното мляко, заради трудното разграждане на лактозата и мазнините. Бях чувал, но засега стомахът ми не дава признаци да не се справя с лактозата, и смятам да си пия млекце... :)

И не щеш ли, някак си попаднах на статия за убиването на нервни клетки. Алкохолът се знае, но и белият хляб беше включен в този списък. По-точно бялото брашно, защото при избелването се получавало някакво токсично за невроните вещество.

Ужас!

Имаше коментар, че е явно, че този токсин влиея в различна степен на различни хора, че убиването става бавно, но в напреднала възраст ефектът му може да се прояви като по-ранно оглупяване.

Искам си всички неврони!!!... :-S
Вече се замислям дали и колко тестени изделия с бяло брашно да ям... :-S

Ужас!

....

I had a lot of fun, reading that caption in Wolverhampton, UK.

My friend told me that white bread is toxic for humans, too, but...
I didn't ask him for details. I guessed it's about fats or because it is harder to digest?


However, recently I faced the truth, when I read papers listing WHITE BREAD as one of the neuron killers, among the alcohol... Precisely, white powder, because the process of making it white produces a substance which is neurotoxic.

Nooooo... I want all of my neurons in tact... :-S

Be aware, fellow researchers!... ;-)
Read More

Sunday, August 10, 2008

// // 2 comments

Smarty – Extendable Framework for Bilingual and Multilingual Comprehension Assistants | Смарти - разширяема архитектура за интелигентни речници

English - see below

По-нова статия и сваляне на програмата:  https://artificial-mind.blogspot.com/2020/06/smarty-comprehension-assistant.html

Смарти - разширяема архитектура за интелигентни речници



( Английско-български интелигентен речник )

Версия Алфа 0.1 НБ




Системни изисквания: Windows XP, .NET 1.1 или 2.0

Изтегли: Smarty 0.1 NB

Zip archive(21 MB)
RAR self-extracting exe(16 MB)


Интересно

- Settings -> Search Endings (търси по суфикси) , напр.:

- Търси и предлага възможни изрази - щракни върху "Expressions". След това ляв бутон върху някоя дума, и в балончето ще излезе възможен израз, а в някои случаи - точният израз. По подразбиране е включено "definition".

Въпроси и отговори

1. Алфа 0.1 НБ?

Алфа - около 3 месеца сериозна работа + дребни доработки, направих системата на една специализация в Улвърхямпън. НБ значи "Няма Балканет". Балканет е речников ресурс, който в съчетание с един друг ресурс - WordNet - позволява понякога да се показва точния смисъл на многозначни думи. Уви, BalkaNet не може да се разпространява безплатно.

2. Бавно зарежда?

Да, трябва да се оправи.


3. Ау, искам по-добър вид на тоя прозорец в средата!

И аз, някой ден ще го оправя.

4. Какви са тия табове Disambiguated-Misambiguated?

Щракай и ще видиш. Повече - в следващи версии.

5. Интелигентен речник? Какво му е интелигентното?! Супер тъп е!

Виж презентациите...

---
It's an Alpha version, consumes a lot of memory, a lot of the interface and the internals need re-engineered.

I need few fellows to do good refactoring and reengineering, some research and improvements of the word-sense disambiguation... access to corpora and web... Some 3-6-9 months work... And I believe "Smarty" might become a really useful tool. :)

But at the moment I can't spend a lot of time for it...


...

Любопитно

Задачата ми беше за 3 месеца сам да създам нещо по-добро от една система на XEROX или от един речник на една друга фирма, Morphologic. :D Чак по-добра - едва ли, но в някои отношения е като тях, и използва един друг подход.

Тази версия съдържа около 9000 реда на С#, разработени основно за около 3 месеца.

...

Малко мразя "Смарти", защото понякога ми взимаше здравето... :-S

Ако се намери с кого да го поправим и подобрим, може да го направим страхотен за 3-6-9 месеца. Сам обаче не знам кога ще му отделя внимание, имам около 1000 проекта...


Smarty – Extendable Framework for Bilingual and Multilingual Comprehension Assistants

"Smarty" was my course work while I was an intern at RIILP in Wolverhampton in 2007. It managed to go to the poster session of LREC 2008.



This is context-sensitive translation, applying POS-tagging, Normalization, Multi-word expressions recognition and even a basic bilingual Word-sense disambiguation, using WordNet with bag-of-word methods and aligned to it lexical database, like BalkaNet or EuroWordNet.



It's an Alpha version, consumes a lot of memory, a lot of the interface and the internals need to be fixed and re-engineered.

I need few fellows to do good refactoring and reengineering, some research and improvements of the word-sense disambiguation... access to corpora and web... Some 3-6-9 months work... And I believe "Smarty" might become a really useful tool. :)

But at the moment unfortunately I can't spend a lot of time on it.
Read More

Monday, August 4, 2008

// // Leave a Comment

Evolution Issue - two compatible superior mutants in the same space and time?

Biological evolution is a cool concept, but it has issues.

That's what I thought about recently:

1. If one new-born creature is evolved to a new superior, "better fitted" type, then it will be genetically incompatible with the type of his or her mother. First of all, will it develop right in his mother, if it is genetically altered new type?

2. Then in order the randomly mutated type to continue to exist, two compatible mutants should be born in a near time-space, and they must bond...

3. On the other hand, if there are very few individuals from the new type, it's very likely that they will not gonna make it. Even if they bond, their children would have the incompatibility problem again - how they would find compatible individuals? And because evolution is "a slow process", the new type probably would be slightly different than the predecessor, thus the new type entities may be not aware that they are not compatible... This would lead to couples which can't have children, and the new type would die for this reason, also. Even if they couple with their own brothers and sisters, this may lead to inferior mutations...

I guess this point is a popular one:

4. Natural Selection sounds good, but it's only valid for one type - individuals from one type become more fitted for particular conditions, but the degree is in the same genetic material of one single type, new types are not created by natural selection.

Issues... :)
Read More