Youtube Youtube Using FFMPEG Experience

Teenagmi

Well-known member
VIP MEMBER
Joined
Jun 1, 2019
Messages
235
The concept of video uniqueness is a complex topic that cannot be addressed with a simple answer. However, I will attempt to explain how it functions and suggest settings you should consider trying.

When it comes to video verification on social networks and video hosting platforms, it can generally be categorized into two main approaches:

  1. Artificial intelligence, machine learning, and live moderators: These methods are typically employed to identify inappropriate content such as explicit material or copyrighted content.
  2. Methods that primarily analyze the technical aspects of the video, including its hash sum.
Personally, I have gained significant experience with the second set of methods, and I will elaborate on them further. The process of altering the video code can be divided into two crucial areas.



  1. The first aspect involves adjusting the recording settings within the video itself. This includes rendering, video bitrate, audio bitrate, frame rate, and constant rate factor (-crf).
  2. The second aspect focuses on making subtle modifications to the video content that are imperceptible to the average viewer. This can include mirroring, adjusting brightness, contrast, adding introductions, or combining random segments of the video, among other techniques.
The number of settings required will depend on the complexity of the task (such as the number of videos to be modified and their original versions). Here are some tips listed in order of increasing complexity, based on the desired outcome and the platform you intend to manipulate:

  1. Start by re-rendering your file. This simple action requires minimal computer resources and time, yet it results in a relatively unique hash sum without altering the video content, as the pixels are overwritten in an arbitrary order.
  2. Set a random bitrate value. Bitrate is a powerful tool in creating unique videos, as it determines the amount of information recorded per second and the overall file size. In my experience, this parameter has geted significant results. The more copies you desire, the wider the range of bitrates you should use (e.g., from 2M to 8M) with increments of 1M for each new video. For example, 2M, 3M, 4M, 5M, 6M, 7M, 8M... Additionally, fractional values can be used, such as setting the video bitrate as "-b:v 6.5M".
  3. Adjust brightness and contrast. This step strengthens the first and second methods by modifying the video content, albeit in a subtle manner that minimally affects the overall quality. The brightness parameter (-vf eq=brightness=) accepts values from -1 to 1. I recommend a range of -0.2 to 0.2, as larger deviations from 0 can noticeably distort the image. In my templates, I generate a random value within the recommended range with a step of 0.02 (e.g., -0.2, -0.18, -0.16, -0.14...). Similarly, the contrast parameter (-vf eq=contrast=) can be adjusted within a range of -1 to 1, with a step of 0.2 (e.g., -1, -0.8, -0.6, -0.4...). Based on my YouTube experience, increments of 0.02 and 0.2 are sufficient to maintain imperceptibility to the human eye while significantly increasing the video's uniqueness from a machine's perspective.
  4. Modify video length and add an intro. Using YouTube as an example, incorporating a randomly prepared intro lasting between 4 to 7 seconds at the beginning of the video can significantly increase the number of unique copies derived from a single file. The intro can consist of a picture, logo, or animated sequence (created using software like After Effects). Creating 3-5 variations of intros and randomly appending them to the main video also alters the video's overall length.
These steps should address 90% of the challenge, resulting in a final approach that can be summarized as follows.


Code:
ffmpeg -i "C:\Users\User\Desktop\Intro.mp4" -acodec copy -vcodec copy -vbsf h264_mp4toannexb -f mpegts "C:\Users\User\Desktop\Intro.ts"
ffmpeg -i "C:\Users\User\Desktop\Video.mp4" -acodec copy -vcodec copy -vbsf h264_mp4toannexb -f mpegts "C:\Users\User\Desktop\Video.ts"
ffmpeg -i "concat:C:\Users\User\Desktop\Intro.ts|C:\Users\User\Desktop\Video.ts" -s 1280x720 -b:v 6.5M -vf eq = brightness = 0.02 -vf eq = contrast = 0.8 "C:\Users\User\Desktop\VideoNew.mp4"

  1. If the previous steps are insufficient for your objectives, the next approach involves generating random values for Frame rate (-r) and Constant Rate Factor (-crf).
For the Frame rate, the value should be selected based on the dynamics of your video, the processing power of your computer, and the desired file size. I would advise against setting it lower than 30, as even for slow slide shows, the visual quality noticeably degrades. Higher frame rates require rendering more frames, which can impact render speed and the overall file size.

The -crf parameter accepts values ranging from 0 to 51, and it determines the video quality. However, our focus lies within the range of 18 to 28. A setting of 18 offers better quality but results in larger file sizes, while 28 gets lower quality but reduces file size. The default value is typically 23.


Code:
ffmpeg -i "C:\Users\User\Desktop\Intro.mp4" -acodec copy -vcodec copy -vbsf h264_mp4toannexb -f mpegts "C:\Users\User\Desktop\Intro.ts"
ffmpeg -i "C:\Users\User\Desktop\Video.mp4" -acodec copy -vcodec copy -vbsf h264_mp4toannexb -f mpegts "C:\Users\User\Desktop\Video.ts"
ffmpeg -i "concat:C:\Users\User\Desktop\Intro.ts|C:\Users\User\Desktop\Video.ts" -s 1280x720 -r 30 -crf 28 -b:v 6.5M -vf eq=brightness=0.02 -vf eq=contrast=0.8 "C:\Users\User\Desktop\VideoNew.mp4"
 
I appreciate your kind words and find it valuable that you found the information I provided on the TikTok forum thread about making unique TikTok videos. Thank you for considering me a legend.
 
Thank you for providing this valuable information! I appreciate it greatly!
 
I have a sense that this information holds significant value, although I haven't fully grasped its meaning yet.
 
How can audio be changed or edited?

I believe that dealing with audio requires specific attention, but I don't have a strong inclination to delve deeply into it. However, I can offer a few suggestions:

  1. One option is to modify the Audio Bitrate (-b:a 320k | 256k | 192k) which can make the audio file distinct but may significantly impact its file size depending on the chosen value.
  2. If I frequently use background soundtracks, I occasionally adjust their volume by +/- 10% and alter their length as a precautionary measure. I cannot ascertain its effectiveness, but I do it for my own peace of mind.
The primary system I use is illustrated in the image below.

Map.jpg

To begin, the process involves randomly selecting content from a pre-prepared main video, along with incorporating a random introduction and random audio.

While I have primarily focused on the video aspect, I haven't devoted as much attention to audio. If someone havees knowledge in this area, I would be delighted to receive new insights and opinions.
 
What type of content are you publishing? And how do you conduct keyword research for it?
 
What type of content are you publishing? And how do you conduct keyword research for it?
Typically, I create short CPA videos. When it comes to conducting keyword research on Google, I rely on the neilpatel service. For YouTube tags, I use vidIQ.
 
If you aim to generate approximately 1000 unique videos, incorporating random intros, outros, and logos can be an effective approach. By using these elements in a random manner, you can create variations that give each video a unique touch. This can help in achieving your goal of generating a large number of distinct videos. Best of luck, and you're very welcome!
 
Can PowerDirector 19 fulfill the requirements for this task?
I haven't encountered PowerDirector 19 before. However, in general, any video editing software with flexible output settings can potentially be suitable for the task. The challenge lies in manually configuring each video personly, which can be time-consuming. On the other hand, ffmpeg offers automation capabilities, making it advantageous in this regard. You mentioned your intention to create another post on this topic, but you're uncertain about the level of detail due to factors such as program usage, computer power, and scalability.
 
Similar threads Most view View more
Back
Top Bottom