Before you start filming, watch

Where to find high quality, well shot and edited news and storytelling videos? Here are a few places:

https://vimeo.com/channels/staffpicks

https://vimeo.com/categories/journalism/videos

https://www.nytimes.com/video

https://edition.cnn.com/videos

https://www.cbsnews.com/60-minutes/

https://www.nytimes.com/by/jonah-m-kessel

https://www.vox.com/authors/johnny-harris

The workflow

• The workflow
  • the proposal, pitch a good story
  • discussion, consider possible scenarios, backup plans
  • preparation, what do you need
  • execution, go, interview, film, video, stills, and sound. Don’t forget names and titles
  • write the story, fact check, proofread, record soundtrack
  • editing, put all elements together
  • watch and repeat
  • delivery, formats, soundcheck

Smartphone journalism

Sound

Visual

Lighting

Software and editing

Editing software is often the first question from many people, Adobe Premiere or Final Cut Pro? For beginners, especially students, it’s easy. If you work on a Windows computer, go with Premiere, if you have a good Macbook, go with either Premiere or Final Cut. All the basic editing rules are universal, just different interface and shortcuts.

Tips before you head out

• Tips for everyone, beginners and pros
  • don’t forget to clean up memory cards, charge the battery, and bring earphones to check the sound quality
  • go for it, but prepare to do it again
  • clean sound beats everything
  • use a tripod if you can
  • keep it rolling, make sure each shot is at least 10 seconds long
• Editing the story
  • start with your best shots
  • keep a good flow, don’t rush, don’t drag
  • leave breathing time between sections, natural sound from the video is often good space fillers
  • the best length is always the shortest length that could tell your story clearly

Learn more about next-level gear/filming

https://www.newsshooter.com/

https://www.youtube.com/c/KaiManWong/videos

https://www.youtube.com/user/filmriot/featured

# over-fit a model to an arbitart target and
# check if the model's parameters are updated

model_optimizer.zero_grad()

# collect all parameters, including sub-modules
# into a list before back propogation and update
# store their sums (easeir than actual values)
param_before = []
for param in model.parameters():
    param_before.append(torch.sum(param.data))

output = self.model(input)

loss = nn.MSELoss()
arbitary_target = torch.tensor(0.4).cuda()

error = loss(output, arbitary_target)

# add to tensorboard for better visualization
writer.add_scalar('loss/error', error, iterations)
writer.add_scalar('loss/out', output.item(), iterations)

error.backward()
model_optimizer.step()

# collect the parameters again after back propogation
param_after = []
for param in model.parameters():
    param_after.append(torch.sum(param.data))

# compare before and after
# if there values are not zeros, the models is updating
# this is helpful when there are many sub-modules
for before, after in zip(param_before, param_after):
    print(f'param change {after - before}')

Description from Beijing Visitor:

“This covered market nestles conveniently between Sanlitun, Sanyuan Qiao and Dongzhimen and has served locals and visitors for over 30 years.

The stallholders are very friendly and happy to pass the time of day, even if you are just curious about where their produce comes from or what exactly it is. Many of them speak a bit of English as they are used to serving the nearby diplomatic community.

For many Western cooks the market is something of a promised land. Supermarket labeling in China can be frustrating, especially when it comes to herbs and spices, many of which are incorrectly translated. Sanyuanli imports everything from marjoram leaves to dill, parsley, rosemary and tumeric, and while this does make some of them a tad expensive, you need look no further if you are on a mission to stock your cupboards.
…
Sanyuanli is worth a visit even if you have no intention of getting behind a stove. Watching the gnomish vendors barter and bargain, their garish winter wear pale in comparison to the deep colours of their vegetables, is a sight that will warm the heart on even the coldest Beijing day.”

Access – how to get to Sanyuanli Wholesale Market

Sanyuanli Wholesale Market
Open 5am – 7pm
West of Sanyuan Dongqiao, Chaoyang
Nearest subway station is Dongzhimen

text by Dave Green, photos by CY Xu

The Issues –

Dynamic Range, as you can tell quite easily from the screen grabs of the footage transcoded using different tools, the 5DtoRGB largely maintained the dynamic range that the GH2 sensor (using a Flow Motion V2 hacked firmware in this case) captured. While the straight output from FCPX or ClipWrap resulted in a “clipped” dynamic range, losing details in dark areas.

Cropping, scroll through the shots you’ll find the straight output and 5DtoRGB output managed to keep the footage uncropped, while ClipWrap crops the image with different codecs.

Don’t be entangled with the reason why this is happening, it’s a long time glitch in QuickTime codec and Apple hasn’t provided an official solution.

The Solution –

Transcoding with the 5DtoRGB before editing.
Lite version free but limits to one file at a time. Batch version costs $49.99 in App Store.

Also this post says ITU-R BT.601 is the correct Decoding Matrix to choose, as comparison, in the 2nd and 3rd pics above, we noticed that the dark area in BT.709 profile are incorrectly exposed and pixelated.

According to 5DtoRGB website —

“5DtoRGB takes a no-compromise approach to quality. 5DtoRGB bypasses QuickTime decoding altogether, works internally at 10 bits and uses your video card’s GPU for its YCbCr to RGB conversion. It also recognizes Canon’s full range 8 bit YCbCr values (0-255), avoiding clipping and the resulting loss of picture information. The resulting files are the absolute highest quality you’ll ever get out of the camera. In fact, you could argue that they’re even better than the camera originals since they’ve undergone high quality chroma smoothing.”

Bonus

Mac AVCHD gamma issues – the fix from EOSHD
An EZ Guide to Hacking the Panasonic GH2 by Sam Mallery
Revisiting the GH2 Hack by Sam Mallery

妞妞，我和爸爸妈妈谢谢你15年的陪伴。
Niuniu, daddy, mummy and I thank you for your 15 years of company.

以后敲门听不到你叫了，以后也没人吃我的剩饭了。
I won’t hear your yowling when I knock on the door, and no one will eat my left-over anymore.

所以，你还是听得懂我说话的，我昨晚在你耳边说悄悄话，“如果太疼了，就走吧。”没想到你真的听懂了。这也成了我跟你说的最后一次悄悄话。
So you understand my words after all. I whispered at your ear last night, “If it’s too painful, just let it go.” I didn’t expect you to understand. This became our last pillow talk.

妞妞，我这辈子应该再也不会养狗了，因为只有你是陪我一起长大的。将来我有了孩子，我会给他们讲你的故事，也会找一只小狗陪他们长大。喜欢的话，你就再陪我的孩子长大吧。
Niuniu, I might not have another dog in my life, as you are the one grew up with me. When I have kids, they will hear your story form me, and I will find a puppy growing up with them. If you like, please come back to our family, grow up with my children again.

谢谢你。
Thank you.

从单位出门到地铁站短短几百米的路上，见过太多人、太多事儿，大多瞥一眼就匆匆走过，少有事情能让我扭头回去找。一只卧在三轮车上的狗，让一路犹犹豫豫到地铁入口的我原路返回。“如果错过，我会后悔的。”

老人坐在路牙子上，旁边的三轮车上有几条脏兮兮的被褥，还有些乱七八糟的破烂。没人会扭头多看一眼，看见了也不会在脑子里多记住一分钟。三轮车上的狗，一条看不出品种的中华田园小猎犬，穿着红色小棉袄，正在车上的一堆破烂里转身找着舒服的姿势卧下。

狗是我回来的原因，但老人似乎才是故事的主角。开始打探老人的故事以前，逗一逗狗总是没错的。可惜等我回来他已经找好姿势睡下了。

老人叫杨站春，唐山人，61岁。

天已经黑了，一个坐在路边的拾荒老人，我居然问他是不是在等人。老人孤身一人，无人可等。没有意外，也没有失望。我问他睡哪，他说他哪都睡。这个三轮车，和这一车破烂，外加这条狗，就是他的全部家当。

他今天啥也没捡上，吃饭钱都没有，身上就一块钱了。说着就把放在衣服内兜里的一块钱掏出来给我看。那会我突然想明白，他不是收破烂，而是拾荒。因为“收”是要有本钱的。

我问他老家人呢，他说就他自己，十岁没娘。没房子，也没地。当初分地分得太少，就没要。他强调自己说的句句实话。是啊老爷子，我干嘛不信你。

老爷子说自己一天没喝水了，因为没捡到矿泉水。天冷了，东西不好捡。那吃饭呢？昨天卖东西挣了十块钱，吃了一份冷炒饼，喝了两瓶啤酒。因为喝啤酒比喝水解渴。而且瓶子还能卖。跟他坐了半个钟头，老爷子给我讲两遍为啥喝啤酒。

“我还花一块钱买了份报纸，我得看报纸啊，不能做违法的事儿。偷啊抢啊那些人都没好下场。”老爷子确实识字儿，他的名字我开始写成了“占”，他说你得加个立。

“我想给狗拍个照片”，“行，你别偷我狗就行。” 养了三个月，我以为是老爷子收养的流浪狗，居然还是他跟人抱来的小崽儿。这也不是老爷子第一条狗了，之前还有一条被偷了。“这狗不怕偷，外人动他他敢咬。谁要偷我狗，看我不打他。”老人给我比划着，“抱来的时候这么小。”

老爷子掏兜，一个塑料袋里是烟丝，还有一个袋里是“今天捡的馒头”。看着老爷子卷烟，我在旁边拍几张照片。“给你也卷一个”，婉言谢绝，虽然我其实挺想试试。

老人的三轮车，是从唐山一路骑来的。“我说了你肯定不信。” 老爷子，我不信自己能从唐山骑过来，我信你。

这么多年就这样过来了，瞪着三轮车捡破烂。老人一直念叨，只要不犯法，干啥都行。“收容我去过，不好受… 偷了抢了那些人都要被抓住的，进监狱，剥夺政治权利终身。”老爷子一边描述着自己亲身的感受，一边掺杂着报纸上学来的词汇给我讲着。

跟老爷子坐了不到半个钟头，今天穿的少，我已经冻透了。把口袋里三十几块零钱留给老爷子，不好意思要，因为他觉得“大家挣点钱都不容易。”我让老爷子买点东西吃，也把狗喂了。他说明天先把自行车链子修了，“不修我一步也骑不动，哪也去不了。”拍拍老爷子叫他多保重，也把狗看好。我出了地铁开车回家，在热的头晕的房间里回忆着那会儿的只言片语，而老爷子那边现在怎样，我想象不到。

我不懂经济学，我不懂社会结构。但我知道中国的贫富差距，并不在于潘石屹和你我之间，其实就在于你我和杨站春老爷子之间。

老爷子说等以后发财了碰见我了请我吃饭。“那咱们就碰吧，希望能再碰见。”

Update: 今天又在路边碰到杨老爷子了，可惜他还没发财，不急。补上一张豆豆的照片。